{ "packId": "infinity-machine", "packName": "The Infinity Machine (book)", "packVersion": "1.0.5", "description": "A knowledge pack based on Sebastian Mallaby's 2026 book about Demis Hassabis, DeepMind, and the quest for superintelligence. Tracks the story from a chess-prodigy childhood in Finchley through Atari and AlphaGo, the founding rivalry with OpenAI, the AlphaFold breakthrough, ChatGPT shock, and the Gemini era.", "author": "Flash Feed", "language": "en", "tagsVocabulary": [ "intro", "hassabis", "biography", "deepmind", "openai", "elixir", "cambridge", "legg", "suleyman", "thiel", "google", "atari", "dqn", "alphago", "alphafold", "jumper", "gemini", "altman", "musk", "safety", "governance", "bletchley", "buchanan", "race", "agi", "transformer", "vocabulary", "people", "milestones", "dates", "places", "policy" ], "items": [ { "id": "fact-sweetness", "shape": "fact", "tags": [ "intro", "hassabis", "milestones" ], "title": "The Sweetness of discovery", "body": "Mallaby opens the book on the moment Demis Hassabis received the 2024 Nobel Prize in Chemistry for AlphaFold. Hassabis calls the feeling of a scientific breakthrough \"the sweetness\" — the brief, private rush when, having struggled against a problem for years, you suddenly see it crack. The whole book is framed as a chase after that feeling.", "illustration": { "imagePrompt": "A scientist in formal attire standing on a ceremony stage, receiving a prestigious medal from a presenter, with an audience seated below.", "imageSearchTerm": "Demis Hassabis Nobel Prize", "alt": "A Nobel Prize gold medal on dark velvet", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/fact-sweetness.jpg", "credit": "Pexels · DS stories", "creditUrl": "https://www.pexels.com/photo/photograph-of-gold-medals-7267543/" }, "factVariant": "image-heavy", "imageCaption": "Years of struggle, then it cracks. Hassabis calls that rush \"the sweetness\" — the whole book is a chase after it.", "studyGuideAnchor": "the-sweetness-of-discovery", "uid": "11si1u01aim3qu" }, { "id": "fact-infinity-machine", "shape": "fact", "tags": [ "intro", "agi", "deepmind" ], "title": "What is the infinity machine?", "body": "The title is Mallaby's metaphor for artificial general intelligence: a single machine that could, in principle, solve any problem a human mind can frame. The book argues that DeepMind's history is the clearest case study of how such a machine is actually being built — not by a moonshot, but by linked breakthroughs in reinforcement learning, deep learning, and now transformers.", "illustration": { "imageSearchTerm": "abstract neural network concept", "imagePrompt": "An abstract network of glowing interconnected nodes forming an endless loop, suggesting a machine capable of solving any problem.", "alt": "What is the infinity machine?", "depictable": false, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:FEQ_July_2018_Brockhampton_(29891515867).jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-infinity-machine.webp" }, "studyGuideAnchor": "what-is-the-infinity-machine", "uid": "h96m6v1be3aft" }, { "id": "pair-author", "shape": "pair", "tags": [ "intro", "vocabulary" ], "sideA": { "modality": "text", "value": "Author of The Infinity Machine" }, "sideB": { "modality": "text", "value": "Sebastian Mallaby" }, "uid": "1tilj05e6cpir" }, { "id": "pair-subject", "shape": "pair", "tags": [ "intro", "hassabis", "vocabulary" ], "sideA": { "modality": "text", "value": "Central subject of the book" }, "sideB": { "modality": "text", "value": "Demis Hassabis" }, "uid": "syd2251dveymz" }, { "id": "mcq-publisher", "shape": "mcq", "tags": [ "intro", "vocabulary" ], "prompt": { "modality": "text", "value": "Which publishing house released The Infinity Machine in 2026?" }, "options": [ { "modality": "text", "value": "Penguin Press" }, { "modality": "text", "value": "Knopf" }, { "modality": "text", "value": "Random House" }, { "modality": "text", "value": "MIT Press" } ], "correctIndex": 0, "uid": "1pn60atyk35m7" }, { "id": "fact-three-strands", "shape": "fact", "tags": [ "intro", "deepmind", "agi" ], "title": "Three strands of the story", "body": "Mallaby braids three storylines: Hassabis's personal arc from chess prodigy to Nobel laureate; the technical lineage from Atari DQN through AlphaGo, AlphaFold and Gemini; and the geopolitics of an arms race between Silicon Valley labs, Google DeepMind, Anthropic, and China. None of the three is intelligible without the other two.", "illustration": { "imageSearchTerm": "braided threads intertwined concept", "imagePrompt": "Three golden threads braided tightly together into one strong cord, symbolizing three storylines converging into one.", "alt": "Three strands of the story", "depictable": false, "url": "https://cdn.recurxive.com/packs/infinity-machine/fact-three-strands.jpg", "credit": "Pexels · Barnabas Sani", "creditUrl": "https://www.pexels.com/photo/close-up-of-frayed-rope-on-green-background-29806627/" }, "studyGuideAnchor": "three-strands-of-the-story", "uid": "c8ino394e3v1" }, { "id": "fact-game-ai", "shape": "fact", "tags": [ "intro", "hassabis", "atari" ], "title": "Why games matter to the plot", "body": "The book keeps returning to games — chess, Diplomacy, Theme Park, Atari, Go, StarCraft. Hassabis's conviction is that games provide a closed environment in which you can measure intelligence against a clear reward, then transfer the learning to the messy world. Every major DeepMind milestone first proved itself in a game.", "illustration": { "imageSearchTerm": "chess board game pieces", "imagePrompt": "A chess board mid-game with pieces in play, representing the classic games AI systems learned to master.", "alt": "Why games matter to the plot", "depictable": true, "credit": "Pexels · Strategic close-up of chess pieces on a board, highlighting the queen in black and white.", "creditUrl": "https://www.pexels.com/photo/black-chess-piece-on-a-chess-board-4800923/", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-game-ai.webp" }, "studyGuideAnchor": "why-games-matter-to-the-plot", "uid": "1qj60ss1d4adjm" }, { "id": "def-agi", "shape": "definition", "tags": [ "intro", "agi", "vocabulary" ], "term": { "modality": "text", "value": "AGI" }, "definition": { "modality": "text", "value": "Artificial general intelligence — a system that can match or surpass human intelligence across the full range of cognitive tasks, not just one narrow domain. Coined by Shane Legg in 2002 in conversations with Marcus Hutter and adopted as DeepMind's founding mission." }, "uid": "1yay55a1fj84gc" }, { "id": "mcq-frame", "shape": "mcq", "tags": [ "intro", "agi" ], "prompt": { "modality": "text", "value": "Mallaby treats DeepMind's mission statement as which kind of project?" }, "options": [ { "modality": "text", "value": "A research lab pursuing AGI as a primary goal" }, { "modality": "text", "value": "A consultancy advising other AI labs" }, { "modality": "text", "value": "A pure software product company" }, { "modality": "text", "value": "A defense contractor for the UK government" } ], "correctIndex": 0, "explanation": "From its 2010 founding, DeepMind explicitly framed itself as an AGI-first organisation — the word was in pitches and incorporation documents.", "uid": "ekrpqk1w3b3vu" }, { "id": "fact-thesis-statement", "shape": "fact", "tags": [ "intro", "agi", "race" ], "title": "The book's thesis in one line", "body": "Mallaby's argument: the most consequential technology of the century is being built by a small group of people whose biographies and rivalries matter — and DeepMind, born of a London poker game and a Singularity Summit pitch, has had outsized influence on what gets built and how.", "illustration": { "imageSearchTerm": "poker table cards chips", "imagePrompt": "A poker table with scattered chips and cards under dim light, symbolizing the personal origins behind a major technological history.", "alt": "The book's thesis in one line", "depictable": false, "credit": "Pexels · Red poker chips stacked with playing cards on a green felt table.", "creditUrl": "https://www.pexels.com/photo/poker-chips-and-playing-cards-on-green-table-29261107/", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-thesis-statement.webp" }, "studyGuideAnchor": "the-book-s-thesis-in-one-line", "curatedDistractors": [ "NeurIPS", "Davos", "TED" ], "uid": "231ikekrxa8s" }, { "id": "fact-birthplace", "shape": "fact", "tags": [ "hassabis", "biography", "places" ], "title": "Born in Finchley, July 1976", "body": "Demis Hassabis was born in July 1976 in Finchley, north London. His mother was Chinese Singaporean; his father was Greek Cypriot. The family lived above the toy shop his father ran. From toddlerhood he was, as Mallaby puts it, \"a born obsessive\" — the kind of child who wanted to understand exactly how every game in the shop worked.", "illustration": { "imagePrompt": "A small toy shop on a north London street with a residential flat visible above it.", "imageSearchTerm": "Finchley London toy shop", "alt": "A quiet residential street in north London", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:81_Church_Lane,_N2.jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-birthplace.webp" }, "factVariant": "image-heavy", "imageCaption": "Demis Hassabis was born in July 1976 in Finchley, north London.", "uid": "1yrivd71eqwtrp" }, { "id": "pair-mother", "shape": "pair", "tags": [ "hassabis", "biography", "vocabulary" ], "sideA": { "modality": "text", "value": "Hassabis's mother's heritage" }, "sideB": { "modality": "text", "value": "Chinese Singaporean" }, "uid": "3s3dyf1grvg59" }, { "id": "pair-father", "shape": "pair", "tags": [ "hassabis", "biography", "vocabulary" ], "sideA": { "modality": "text", "value": "Hassabis's father's heritage" }, "sideB": { "modality": "text", "value": "Greek Cypriot" }, "uid": "180xo4oqgpqx6" }, { "id": "fact-chess", "shape": "fact", "tags": [ "hassabis", "biography" ], "title": "Chess master at 13", "body": "Hassabis learned chess at four after watching his father play, and within months was beating adults. He reached master strength by age 13, briefly the second-highest-rated child player in the world for his age. The chess world taught him how to think about adversarial systems — but also showed him that humans had cognitive ceilings he could already feel pressing in.", "illustration": { "imageSearchTerm": "Chess master at 13", "imagePrompt": "Chess master at 13. Hassabis learned chess at four after watching his father play, and within months was beating adults. He reached master strength by age 13, briefly the second-highest-rated child pl", "alt": "Chess master at 13", "credit": "Pexels · Vlada Karpovich", "creditUrl": "https://www.pexels.com/photo/boy-in-white-and-blue-plaid-button-up-shirt-6115014/", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-chess.webp" }, "uid": "wv84h31bh779" }, { "id": "numeric-master-age", "shape": "numeric", "tags": [ "hassabis", "biography", "milestones" ], "prompt": { "modality": "text", "value": "Age at which Hassabis reached chess master strength" }, "value": 13, "unit": "years", "uid": "j5uqu91q7gxzf" }, { "id": "fact-zx-spectrum", "shape": "fact", "tags": [ "hassabis", "biography" ], "title": "ZX Spectrum and self-taught code", "body": "With chess winnings, eight-year-old Demis bought a ZX Spectrum home computer. He devoured manuals and taught himself BASIC and machine code. He coded a neural network long before he understood the theory — he simply wanted to make the computer play games against itself. The mental thread from those early experiments runs directly to AlphaZero.", "illustration": { "imageSearchTerm": "ZX Spectrum and self-taught code", "imagePrompt": "ZX Spectrum and self-taught code. With chess winnings, eight-year-old Demis bought a ZX Spectrum home computer. He devoured manuals and taught himself BASIC and machine code. He coded a neural network long before h", "alt": "ZX Spectrum and self-taught code", "credit": "Pexels · William Warby", "creditUrl": "https://www.pexels.com/photo/black-vintage-keyboard-19599876/", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-zx-spectrum.webp" }, "uid": "1k0gow4dnyu4e" }, { "id": "pair-bullfrog", "shape": "pair", "tags": [ "hassabis", "biography", "vocabulary" ], "sideA": { "modality": "text", "value": "Peter Molyneux's game studio" }, "sideB": { "modality": "text", "value": "Bullfrog" }, "uid": "1pu0przde5j4p" }, { "id": "mcq-theme-park", "shape": "mcq", "tags": [ "hassabis", "biography" ], "prompt": { "modality": "text", "value": "Which Bullfrog game did the teenage Hassabis co-design?" }, "options": [ { "modality": "text", "value": "Populous" }, { "modality": "text", "value": "Theme Park" }, { "modality": "text", "value": "Black & White" }, { "modality": "text", "value": "Dungeon Keeper" } ], "correctIndex": 1, "uid": "llgdq73nxf7h" }, { "id": "fact-geb", "shape": "fact", "tags": [ "hassabis", "biography" ], "title": "Gödel, Escher, Bach", "body": "Mallaby singles out a single book — Douglas Hofstadter's Gödel, Escher, Bach — as the most important text of Hassabis's teenage years. Its braided account of self-reference, formal systems and the strange loops of mind gave Hassabis his first sense that intelligence might be something you could build out of layered patterns rather than something mystical.", "illustration": { "imageSearchTerm": "Gödel Escher Bach book", "imagePrompt": "An open hardcover book lying on a wooden desk with abstract geometric line-drawing patterns visible on its pages.", "alt": "Cover of the book Gödel, Escher, Bach", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Carl_G%C3%B6del_Perchtoldsdorf.jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-geb.webp" }, "uid": "dbvs4kmhgvle" }, { "id": "fact-chess-ceiling", "shape": "fact", "tags": [ "hassabis", "biography" ], "title": "Quitting tournament chess", "body": "Hassabis stopped playing competitive chess in his late teens. The trigger, he later told Mallaby, was watching adults sacrifice everything to be slightly stronger than each other on a sixty-four-square board: \"I started to wonder what else I could do with all that thinking.\" The chess habit never left him — but he wanted a bigger board.", "illustration": { "imageSearchTerm": "chess tournament players board", "imagePrompt": "Two competitors seated at a chess tournament table, deep in concentration over a mid-game board.", "alt": "Quitting tournament chess", "depictable": true, "credit": "Pexels · Indoor chess tournament with players, focusing on a young adult participant.", "creditUrl": "https://www.pexels.com/photo/chess-tournament-venue-with-participants-28397730/", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-chess-ceiling.webp" }, "uid": "1l2mlu7r5fwwx" }, { "id": "fact-pdp-1991", "shape": "fact", "tags": [ "hassabis", "biography" ], "title": "A first brush with neural nets", "body": "In 1991, still a teenager, Hassabis read about the PDP (Parallel Distributed Processing) movement and experimented with backprop on his home machine. He could feel that the technique was important but underpowered. Twenty years later, when graphics cards made it powerful, he was already mentally rehearsed.", "illustration": { "imageSearchTerm": "1990s home computer desk", "imagePrompt": "An early 1990s home computer with a bulky CRT monitor sitting on a teenager's bedroom desk.", "alt": "A first brush with neural nets", "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:1990s_decade_montage_3.0.jpg", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-pdp-1991.webp" }, "uid": "y6sazt12mao7z" }, { "id": "fact-cambridge", "shape": "fact", "tags": [ "hassabis", "cambridge", "biography" ], "title": "Cambridge: Queens' College", "body": "In 1994 Hassabis went up to Queens' College, Cambridge to read computer science. He was three years older than most freshers because of his time at Bullfrog. Cambridge was where, for the first time, his own intellect did not feel exceptional — and where he met collaborators he would still be working with twenty-five years later.", "illustration": { "imagePrompt": "A wooden bridge over the River Cam at a historic Cambridge college, with old stone buildings behind it.", "imageSearchTerm": "Queens College Cambridge bridge", "alt": "Queens' College and the Mathematical Bridge at Cambridge", "depictable": true, "credit": "John Welford  · CC BY-SA 2.0", "creditUrl": "https://commons.wikimedia.org/wiki/File:The_Mathematical_Bridge,_Queens%27_College,_Cambridge_-_geograph.org.uk_-_4560544.jpg", "subject": "The Mathematical Bridge: timber tangent-and-radial truss over the River Cam, a punt with passengers passing beneath, Queens' College red-brick riverside range with Gothic windows and a lantern on the right; overcast sky", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-cambridge.webp" }, "factVariant": "image-heavy", "imageCaption": "In 1994 Hassabis went up to Queens' College, Cambridge to read computer science.", "studyGuideAnchor": "cambridge-queens-college", "uid": "oicl0ekaloy" }, { "id": "pair-david-silver", "shape": "pair", "tags": [ "cambridge", "people", "vocabulary" ], "sideA": { "modality": "text", "value": "Cambridge friend and lifelong DeepMind collaborator" }, "sideB": { "modality": "text", "value": "David Silver" }, "uid": "1m5hczpv2os1v" }, { "id": "fact-silver-meeting", "shape": "fact", "tags": [ "cambridge", "people" ], "title": "Meeting David Silver", "body": "In his first weeks at Cambridge, Hassabis met another freshman, David Silver, in the JCR. They bonded over Go, board games and a shared sense that the computer science syllabus was not asking the deep philosophical questions. Silver would later lead the AlphaGo project and, decades later, help design AlphaFold's reinforcement learning.", "illustration": { "imageSearchTerm": "students playing Go game", "imagePrompt": "Two university students in a common room, deep in a game of Go on a wooden board between them.", "alt": "Meeting David Silver", "depictable": true, "credit": "Pexels · Aerial view of people playing Go outdoors, showcasing a casual game setup on a grassy lawn.", "creditUrl": "https://www.pexels.com/photo/people-playing-go-outdoors-on-grass-lawn-32599544/", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-silver-meeting.webp" }, "studyGuideAnchor": "meeting-david-silver", "uid": "fk71gk1avggtu" }, { "id": "fact-deep-philosophical", "shape": "fact", "tags": [ "cambridge", "hassabis", "agi" ], "title": "The deep philosophical questions", "body": "Mallaby titles a chapter after Hassabis's phrase. Cambridge convinced Hassabis that the most interesting question in computer science was not the next algorithm but the nature of mind itself. He resolved that, after a few years making money in games, he would return to neuroscience and AI for the long haul.", "illustration": { "imageSearchTerm": "brain neural network illustration", "imagePrompt": "A human brain rendered as glowing interconnected circuitry, blending biological form with abstract computation.", "alt": "The deep philosophical questions", "depictable": false, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Aires_c%C3%A9r%C3%A9brales_de_la_motricit%C3%A9_volontaire.jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-deep-philosophical.webp" }, "studyGuideAnchor": "the-deep-philosophical-questions", "uid": "1ryy3gh2y850v" }, { "id": "fact-elixir-founded", "shape": "fact", "tags": [ "elixir", "milestones" ], "title": "Elixir Studios founded, 7 July 1998", "body": "On 7 July 1998, aged 22, Hassabis founded Elixir Studios in London. The plan was to build the kind of ambitious AI-driven games he had imagined since Bullfrog. Elixir would produce Republic: The Revolution and Evil Genius — neither a financial smash, but both prized for their AI ambition.", "illustration": { "imageSearchTerm": "1990s video game studio", "imagePrompt": "A late-1990s video game development studio office with programmers working at CRT monitors surrounded by concept art.", "alt": "Elixir Studios founded, 7 July 1998", "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:1990s_montage.png", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-elixir-founded.webp" }, "studyGuideAnchor": "elixir-studios-founded-7-july-1998", "uid": "wzp9ffbpzyxt" }, { "id": "numeric-elixir-year", "shape": "numeric", "tags": [ "elixir", "milestones", "dates" ], "prompt": { "modality": "text", "value": "Year Hassabis founded Elixir Studios" }, "value": 1998, "unit": "year", "uid": "iifhkrurcwjt" }, { "id": "pair-elixir-game", "shape": "pair", "tags": [ "elixir", "vocabulary" ], "sideA": { "modality": "text", "value": "Elixir Studios' satirical spy-villain game" }, "sideB": { "modality": "text", "value": "Evil Genius" }, "uid": "1bt487n4dhred" }, { "id": "mcq-republic", "shape": "mcq", "tags": [ "elixir" ], "prompt": { "modality": "text", "value": "What was Elixir's flagship political simulation, modelling thousands of citizens with their own goals?" }, "options": [ { "modality": "text", "value": "Tropico" }, { "modality": "text", "value": "Republic: The Revolution" }, { "modality": "text", "value": "Black & White" }, { "modality": "text", "value": "Civilization V" } ], "correctIndex": 1, "explanation": "Republic: The Revolution was a 2003 release with an enormously ambitious AI back-end that overshot the era's hardware. Critics admired the ambition but found play sluggish.", "uid": "1b0mj81qrmnj6" }, { "id": "fact-elixir-end", "shape": "fact", "tags": [ "elixir", "biography" ], "title": "The end of Elixir", "body": "Elixir struggled commercially. By 2005, with the studio winding down, Hassabis sold the rights to the games it had completed and shut Elixir's gates. He returned to academia, enrolling at UCL to do a PhD in cognitive neuroscience — looking, specifically, at the brain mechanism behind memory and imagination.", "illustration": { "imageSearchTerm": "University College London building", "imagePrompt": "The neoclassical facade of a grand university building in London under a grey sky.", "alt": "The end of Elixir", "depictable": true, "credit": "Pexels · Close-up of King's College London flag on a modern architectural building facade.", "creditUrl": "https://www.pexels.com/photo/flag-of-kings-college-in-london-18504448/", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-elixir-end.webp" }, "studyGuideAnchor": "the-end-of-elixir", "uid": "16xy4dj52n1kd" }, { "id": "fact-ucl-phd", "shape": "fact", "tags": [ "hassabis", "biography" ], "title": "PhD in cognitive neuroscience", "body": "Hassabis's UCL doctorate, with Eleanor Maguire, found that patients with hippocampal damage could not imagine future scenes — the same brain region that stores memories also constructs hypothetical futures. The paper made the cover of Science in 2007 and gave Hassabis a working theory of the brain that would shape DeepMind's design.", "illustration": { "imageSearchTerm": "brain MRI scan hippocampus", "imagePrompt": "A brain MRI scan image highlighting the hippocampus region at the center of the brain.", "alt": "PhD in cognitive neuroscience", "depictable": true, "credit": "Pexels · A doctor's hand points to a brain MRI scan on a lightbox, illustrating medical diagnosis.", "creditUrl": "https://www.pexels.com/photo/finger-pointing-to-a-brain-scan-4226119/", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-ucl-phd.webp" }, "studyGuideAnchor": "phd-in-cognitive-neuroscience", "uid": "1n1c71ztb6tat" }, { "id": "fact-elixir-lesson", "shape": "fact", "tags": [ "elixir", "biography" ], "title": "The lesson of Elixir", "body": "Hassabis later said Elixir taught him that ambition without focus is fatal. Trying to build six games at once and reinvent AI for each one is how studios die. At DeepMind he would impose ruthless prioritisation: a single, well-resourced strike team on each grand-challenge problem.", "illustration": { "imageSearchTerm": "single spotlight path focus", "imagePrompt": "A single spotlight cutting through darkness to illuminate one clear path, while several dim side-routes fade into shadow.", "alt": "The lesson of Elixir", "depictable": false, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Single_bananas_in_a_bunch.jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-elixir-lesson.webp" }, "studyGuideAnchor": "the-lesson-of-elixir", "uid": "vjkx2f1b002rt" }, { "id": "fact-shane-legg", "shape": "fact", "tags": [ "legg", "biography" ], "title": "Shane Legg — coiner of \"AGI\"", "body": "Born in New Zealand in 1973, Shane Legg drifted through cognitive science, finance and machine-learning before landing at IDSIA in Lugano under Marcus Hutter and Jürgen Schmidhuber. While brainstorming a book title with Hutter around 2002, he proposed \"Artificial General Intelligence\". The phrase stuck. Legg would later cofound DeepMind.", "illustration": { "imageSearchTerm": "Shane Legg portrait", "imagePrompt": "A machine-learning researcher sitting thoughtfully at a desk in a modern research office.", "alt": "Shane Legg — coiner of \"AGI\"", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-shane-legg.webp", "credit": "Pexels · File:Shane Cotton (cropped).jpg", "creditUrl": "https://commons.wikimedia.org/wiki/File:Shane_Cotton_(cropped).jpg" }, "uid": "r0kle4mj934y" }, { "id": "pair-legg-term", "shape": "pair", "tags": [ "legg", "agi", "vocabulary" ], "sideA": { "modality": "text", "value": "Coined the term \"AGI\"" }, "sideB": { "modality": "text", "value": "Shane Legg" }, "uid": "ktj3lt14s5xw3" }, { "id": "pair-idsia", "shape": "pair", "tags": [ "legg", "places", "vocabulary" ], "sideA": { "modality": "text", "value": "Swiss AI lab where Legg studied with Hutter" }, "sideB": { "modality": "text", "value": "IDSIA" }, "uid": "8rqbt14ajpj3" }, { "id": "fact-halloween", "shape": "fact", "tags": [ "legg", "milestones" ], "title": "The Halloween Scenario", "body": "On 31 October 2009 Legg gave a Gatsby Unit lecture at UCL titled \"The Halloween Scenario\". His thesis: powerful AI was no longer a thought experiment, and serious safety planning had to begin now. In the audience was Hassabis. They started talking that night and never really stopped.", "illustration": { "imageSearchTerm": "university lecture hall audience", "imagePrompt": "A university lecture hall with a speaker at the podium and an attentive audience seated in rows.", "alt": "The Halloween Scenario", "credit": "Pexels · Students of various backgrounds attentively listening in a university lecture hall.", "creditUrl": "https://www.pexels.com/photo/students-sitting-on-their-chairs-8197505/", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-halloween.webp" }, "uid": "jrczcmoyx2r8" }, { "id": "fact-suleyman-bio", "shape": "fact", "tags": [ "suleyman", "biography" ], "title": "Mustafa \"Moose\" Suleyman", "body": "Born in August 1984 to a Syrian taxi-driver father and English mother in north London, Mustafa Suleyman dropped out of Oxford to start a Muslim youth helpline. By his mid-twenties he was a conflict-resolution consultant working in Copenhagen on climate diplomacy. He met Hassabis through Demis's brother George — they had been at school together — and was recruited as DeepMind's third cofounder.", "illustration": { "imagePrompt": "A tech executive in a smart casual blazer speaking during an interview or panel discussion.", "imageSearchTerm": "Mustafa Suleyman portrait", "alt": "A young entrepreneur speaking at a conference", "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:III._Selim_ve_IV._Mustafa%27n%C4%B1n_elyaz%C4%B1lar%C4%B1.png", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-suleyman-bio.webp" }, "factVariant": "image-heavy", "imageCaption": "An Oxford dropout who ran a youth helpline became DeepMind's third cofounder.", "uid": "w9pduu771sk" }, { "id": "pair-moose", "shape": "pair", "tags": [ "suleyman", "vocabulary" ], "sideA": { "modality": "text", "value": "Suleyman's nickname inside DeepMind" }, "sideB": { "modality": "text", "value": "Moose" }, "uid": "qovypi1xvi6s8" }, { "id": "fact-cofounders", "shape": "fact", "tags": [ "deepmind", "people" ], "title": "Three cofounders, three temperaments", "body": "Mallaby's central triangle: Hassabis as the obsessive games-and-neuroscience polymath; Legg as the cerebral safety-aware theorist; Suleyman as the relentless operator who could pitch a billionaire at midnight. Each of the three would later run a different kind of AI organisation.", "illustration": { "imageSearchTerm": "DeepMind founders group photo", "imagePrompt": "Three young entrepreneurs standing together in a small office, the early founding team of a startup.", "alt": "Three cofounders, three temperaments", "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:SD_2025_-_Jeff_Dean_02.jpg", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-cofounders.webp" }, "uid": "vg7z721vjby7k" }, { "id": "mcq-third-cofounder", "shape": "mcq", "tags": [ "deepmind", "people" ], "prompt": { "modality": "text", "value": "Who was the third cofounder of DeepMind alongside Hassabis and Legg?" }, "options": [ { "modality": "text", "value": "Mustafa Suleyman" }, { "modality": "text", "value": "Geoffrey Hinton" }, { "modality": "text", "value": "Ilya Sutskever" }, { "modality": "text", "value": "David Silver" } ], "correctIndex": 0, "uid": "1bw8wqtagx9mn" }, { "id": "fact-vic-poker", "shape": "fact", "tags": [ "deepmind", "thiel", "milestones" ], "title": "Poker at the Vic, 15 July 2010", "body": "On 15 July 2010, Hassabis, Legg and Suleyman were playing poker at the Vic Casino in west London — they were already incubating DeepMind. By the end of the night they had agreed: the first major fundraising target would be Peter Thiel. Five weeks later they would corner him in California.", "illustration": { "imageSearchTerm": "poker table casino night", "imagePrompt": "A dimly lit casino poker table with cards and chips mid-game, evoking a late-night high-stakes conversation.", "alt": "Poker at the Vic, 15 July 2010", "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Charlie_Carrel_-_Poker_(Double_AA%27s).jpg", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-vic-poker.webp" }, "uid": "305kb495ajpu" }, { "id": "fact-singularity-summit", "shape": "fact", "tags": [ "deepmind", "thiel", "milestones" ], "title": "Singularity Summit, 14 August 2010", "body": "At the 14 August 2010 Singularity Summit in San Francisco, the trio targeted Peter Thiel. Suleyman engineered an introduction. Hassabis pitched a research lab whose explicit mission would be to build AGI safely. Thiel listened, said little, and invited the team to his Palo Alto house for breakfast the next day.", "illustration": { "imageSearchTerm": "tech conference stage speaker", "imagePrompt": "A speaker presenting on stage at a technology conference, with an audience seated below.", "alt": "Singularity Summit, 14 August 2010", "credit": "Pexels · A speaker engages the audience during a tech conference, discussing future trends.", "creditUrl": "https://www.pexels.com/photo/keynote-speaker-at-technology-conference-34774347/", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-singularity-summit.webp" }, "uid": "vwyblq1qyb27w" }, { "id": "fact-thiel-pitch", "shape": "fact", "tags": [ "thiel", "deepmind" ], "title": "\"Demis's destiny\"", "body": "At the Palo Alto breakfast, Thiel — by Mallaby's account — was charmed less by the technical plan than by Hassabis himself. Walking the trio back to their car, he muttered to Luke Nosek that this was \"Demis's destiny.\" Founders Fund led the seed round with $2 million; a $3.7 million Series A followed.", "illustration": { "imageSearchTerm": "Peter Thiel portrait", "imagePrompt": "A venture investor in a blazer standing at a breakfast table in a sunlit California home, mid-conversation.", "alt": "\"Demis's destiny\"", "depictable": true, "credit": "david.orban (Openverse) · by 2.0", "creditUrl": "https://www.flickr.com/photos/44124368329@N01/1351681091", "subject": "close head-and-shoulders shot of a dark-haired man with light eyes in a striped shirt and lanyard, mouth open mid-speech, a woman partly visible behind at left; consistent with Peter Thiel c. 2007", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-thiel-pitch.webp" }, "factVariant": "image-heavy", "imageCaption": "At the Palo Alto breakfast, Thiel — by Mallaby's account — was charmed less by the technical plan than by Hassabis himself.", "uid": "1atb30n1h4l7fd" }, { "id": "fact-deepmind-name", "shape": "fact", "tags": [ "deepmind", "vocabulary" ], "title": "The name DeepMind", "body": "The company's first working name was Solaria, after the Asimov novel. Legg argued for something simpler. \"Deep\" pointed to deep learning, the technique they would bet on; \"Mind\" stated the goal. The name was registered in late 2010 and the company filed in London the following year.", "illustration": { "imageSearchTerm": "brainstorm whiteboard company names", "imagePrompt": "Two overlapping abstract shapes, one evoking depth and one evoking a human mind, merging into a single emblem.", "alt": "The name DeepMind", "depictable": false, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:04_Jubalts.jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-deepmind-name.webp" }, "studyGuideAnchor": "the-name-deepmind", "uid": "1o34ion17ki40d" }, { "id": "fact-google-acquisition", "shape": "fact", "tags": [ "google", "deepmind", "milestones" ], "title": "Google buys DeepMind, January 2014", "body": "Larry Page and Sergey Brin had been circling DeepMind since 2012. After a year of negotiations, in January 2014 Google paid roughly £400 million for the still-revenueless lab — at the time Europe's largest AI acquisition. A condition Hassabis insisted on: an independent ethics board to oversee any potentially dangerous applications.", "illustration": { "imagePrompt": "The glass and steel exterior of a modern tech company campus building, photographed under an overcast sky.", "imageSearchTerm": "Google headquarters campus building", "alt": "The Google logo lit up on a corporate building", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:GSoC_Book_Jotter.jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-google-acquisition.webp" }, "factVariant": "image-heavy", "imageCaption": "Google pays ~£400M for a revenueless lab — Europe's biggest AI buy. The catch: an independent ethics board.", "studyGuideAnchor": "google-buys-deepmind-january-2014", "uid": "18hpqj0x76hr2" }, { "id": "numeric-acquisition-price", "shape": "numeric", "tags": [ "google", "deepmind", "milestones" ], "prompt": { "modality": "text", "value": "Approximate price (£ million) Google paid for DeepMind in 2014" }, "value": 400, "unit": "£ million", "tolerance": 50, "uid": "upln5f13kyxz9" }, { "id": "fact-musk-page", "shape": "fact", "tags": [ "google", "musk" ], "title": "Musk vs Page on the lawn", "body": "On 8 October 2012, Elon Musk dined at Larry Page's home. The conversation turned to AI. Musk warned that an advanced AI could become \"a digital god.\" Page replied that the species who built it would be its rightful heirs. Mallaby treats the night as the moment the AI alliance between Musk and Google ruptured — and the seed of OpenAI was planted.", "illustration": { "imageSearchTerm": "Elon Musk Larry Page", "imagePrompt": "Two men in casual clothing standing on a backyard lawn at dusk, deep in conversation.", "alt": "Google co-founder Larry Page speaking", "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Elon_-_Ulricehamn_-_20250927_-_02.jpg", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-musk-page.webp" }, "factVariant": "image-heavy", "imageCaption": "Musk warned it could become \"a digital god.\" Page said its builders would be its heirs. One dinner, and OpenAI was born.", "studyGuideAnchor": "musk-vs-page-on-the-lawn", "uid": "l2b0g11df69n3" }, { "id": "fact-ethics-board", "shape": "fact", "tags": [ "deepmind", "governance" ], "title": "The (largely silent) ethics board", "body": "As a condition of the Google deal, DeepMind insisted on an Ethics and Safety Review Board: external members who would have to be informed before any AGI-relevant breakthrough. In practice, Mallaby reports, the board met rarely and was eventually folded into Google's broader processes. It was a victory of optics over governance — but it set a template the industry has been arguing about ever since.", "illustration": { "imageSearchTerm": "empty corporate boardroom table", "imagePrompt": "An ornate, empty chair at the head of an otherwise vacant boardroom table, symbolizing an oversight body that rarely convened.", "alt": "The (largely silent) ethics board", "depictable": false, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:2015_Vauxhall_Vivaro,_rear_interior.jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-ethics-board.webp" }, "studyGuideAnchor": "the-largely-silent-ethics-board", "uid": "heuzdd17rx577" }, { "id": "fact-atari-team", "shape": "fact", "tags": [ "atari", "dqn", "deepmind" ], "title": "The Atari strike team", "body": "DeepMind's first publicly recognised milestone came from a tiny team led by Vlad Mnih, a Toronto-Alberta-trained reinforcement-learning specialist. The challenge: build a single neural network that could learn to play any of the classic Atari 2600 games from raw screen pixels and a score signal — no game-specific code.", "illustration": { "imageSearchTerm": "Atari 2600 game screen", "imagePrompt": "A vintage game console connected to a CRT TV displaying a classic arcade game mid-play.", "alt": "The Atari strike team", "credit": "Pexels · Close-up of a retro arcade game 'Game Over' screen in green, evoking classic gaming nostalgia.", "creditUrl": "https://www.pexels.com/photo/retro-arcade-game-over-screen-in-green-35247741/", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-atari-team.webp" }, "studyGuideAnchor": "the-atari-strike-team", "uid": "a6aqwe15xbd4" }, { "id": "pair-mnih", "shape": "pair", "tags": [ "atari", "dqn", "people", "vocabulary" ], "sideA": { "modality": "text", "value": "Lead researcher on the Atari DQN agent" }, "sideB": { "modality": "text", "value": "Vlad Mnih" }, "uid": "gblfpx1mnqryf" }, { "id": "def-dqn", "shape": "definition", "tags": [ "dqn", "atari", "vocabulary" ], "term": { "modality": "text", "value": "DQN (Deep Q-Network)" }, "definition": { "modality": "text", "value": "A reinforcement-learning architecture that pairs Q-learning — predicting the long-term value of each possible action — with a deep convolutional network that reads raw pixels. DQN, with experience replay for stable training, was DeepMind's first headline result and the model for everything that followed." }, "uid": "3oi1bt1ayx403" }, { "id": "def-q-learning", "shape": "definition", "tags": [ "dqn", "vocabulary" ], "term": { "modality": "text", "value": "Q-learning" }, "definition": { "modality": "text", "value": "A reinforcement-learning algorithm in which an agent learns the expected long-term reward of taking each action from each state. The Q stands for the quality of an action — and the trick is iteratively updating the estimate as new rewards arrive." }, "uid": "kmojqk2qhkt2" }, { "id": "fact-experience-replay", "shape": "fact", "tags": [ "dqn", "atari" ], "title": "Why experience replay mattered", "body": "Neural networks hate seeing highly correlated training examples in sequence — they overfit to whatever's in front of them. The DQN insight was to store the agent's experiences in a memory buffer and train the network on random samples from that buffer. This decorrelated the data and made deep RL stable enough to actually learn.", "illustration": { "imageSearchTerm": "abstract data memory buffer", "imagePrompt": "A stream of glowing data fragments flowing into a swirling holding pool, then being drawn out again in scrambled order.", "alt": "Why experience replay mattered", "depictable": false, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:FEQ_July_2018_Brockhampton_(43017650620)_(cropped).jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-experience-replay.webp" }, "studyGuideAnchor": "why-experience-replay-mattered", "uid": "srgr5ab51l9o" }, { "id": "fact-pong-day", "shape": "fact", "tags": [ "atari", "milestones" ], "title": "The day Pong clicked", "body": "On 7 July 2013 — Wimbledon final day — the team watched the DQN agent learn to play Pong from scratch. After a few hours, the bot started to put the paddle in the right place. By that evening it was beating the built-in opponent. Several engineers wept. The intellectual lineage to AlphaGo, AlphaFold and Gemini all run through that afternoon.", "illustration": { "imageSearchTerm": "Pong video game screen", "imagePrompt": "A classic Pong video game screen showing two paddles and a ball mid-rally on a black background.", "alt": "An original Pong arcade cabinet", "depictable": true, "credit": "Unsplash · P. L. · Unsplash License", "creditUrl": "https://unsplash.com/photos/a-yellow-arcade-machine-sitting-on-top-of-a-wooden-floor-U4mxkp2HQGU", "subject": "A single yellow-and-woodgrain Atari Pong arcade cabinet photographed front-on: 'PONG' on the marquee, screen showing two paddles, ball, dotted centre line and score 11-8, two-dial PLAYER 1 / PLAYER 2 control panel marked ATARI INC", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-pong-day.webp" }, "studyGuideAnchor": "the-day-pong-clicked", "uid": "16cfjgzzffm9" }, { "id": "fact-49-games", "shape": "fact", "tags": [ "dqn", "atari" ], "title": "Forty-nine games, one network", "body": "By the time DQN was published in Nature in February 2015, a single network architecture — same weights re-trained from scratch — had reached human-level or better on 29 of 49 Atari games, with no hand-engineered features per game. It was the first compelling demonstration that one general algorithm could master many distinct environments.", "illustration": { "imagePrompt": "A collection of colorful vintage game cartridges spread out next to a retro console.", "imageSearchTerm": "Atari 2600 game cartridges", "alt": "An Atari 2600 console and CRT television", "depictable": true, "credit": "Pexels · Close-up of vintage Atari game cartridges for Missile Command, Pac-Man, and Defender.", "creditUrl": "https://www.pexels.com/photo/game-cartridges-1373100/", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-49-games.webp" }, "factVariant": "image-heavy", "imageCaption": "One network, same weights, human-level or better on 29 of 49 Atari games — with no game-specific tuning.", "studyGuideAnchor": "forty-nine-games-one-network", "uid": "wve49e1aitje8" }, { "id": "fact-brin-go", "shape": "fact", "tags": [ "alphago", "google", "milestones" ], "title": "Sergey Brin and the Go problem", "body": "On a hallway walk at Google in May 2014, Sergey Brin asked Hassabis what hard problem DeepMind would tackle next. Hassabis answered: \"Go.\" Brin was sceptical — the game's complexity had defeated AI for decades — but he provided the resources. The AlphaGo project began that summer.", "illustration": { "imageSearchTerm": "Sergey Brin portrait", "imagePrompt": "A tech company co-founder walking down a modern office hallway in conversation with a colleague.", "alt": "Portrait of Google co-founder Sergey Brin", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Carte_commune_Sergey_2017.png", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-brin-go.webp" }, "studyGuideAnchor": "sergey-brin-and-the-go-problem", "uid": "1sr93b3njyy5" }, { "id": "pair-aja", "shape": "pair", "tags": [ "alphago", "people", "vocabulary" ], "sideA": { "modality": "text", "value": "Engineer who built AlphaGo's first prototype" }, "sideB": { "modality": "text", "value": "Aja Huang" }, "uid": "11y229p5jl3az" }, { "id": "fact-fan-hui", "shape": "fact", "tags": [ "alphago", "milestones" ], "title": "Fan Hui, 5–0, October 2015", "body": "In October 2015, AlphaGo played the European Go champion Fan Hui in a closed-door match. The result, 5–0 to AlphaGo, was kept under embargo until the January 2016 Nature cover paper. Fan Hui was so impressed by the experience that he joined DeepMind as an analyst to help prepare for Lee Sedol.", "illustration": { "imageSearchTerm": "Go match players board", "imagePrompt": "Two players seated at a Go board mid-tournament match, black and white stones arranged across the grid.", "alt": "Fan Hui, 5–0, October 2015", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Go_Nagai_20080704_Japan_Expo_04.jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-fan-hui.webp" }, "studyGuideAnchor": "fan-hui-5-0-october-2015", "uid": "ukco1sf30tuu" }, { "id": "fact-lee-sedol", "shape": "fact", "tags": [ "alphago", "milestones" ], "title": "Lee Sedol, Seoul, March 2016", "body": "Over a week in Seoul in March 2016, AlphaGo beat the legendary Korean champion Lee Sedol 4–1 in front of a global television audience. Lee won game 4 by finding a move AlphaGo had not foreseen — a moment of human glory in an otherwise grim defeat. The match made artificial intelligence front-page news worldwide.", "illustration": { "imagePrompt": "A champion Go player seated in concentration at a tournament board during a televised match.", "imageSearchTerm": "Lee Sedol Go match", "alt": "A Go board mid-game with black and white stones", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/fact-lee-sedol.jpg", "credit": "Pexels · Audrey B", "creditUrl": "https://www.pexels.com/photo/minimalist-go-board-game-setup-with-plant-37403742/" }, "factVariant": "image-heavy", "imageCaption": "Seoul, March 2016: AlphaGo beats the world's best Go player 4–1 on live TV. AI hits the front page everywhere.", "studyGuideAnchor": "lee-sedol-seoul-march-2016", "uid": "1irbgduix49hw" }, { "id": "fact-move-37", "shape": "fact", "tags": [ "alphago", "milestones" ], "title": "Move 37", "body": "In game two, on move 37, AlphaGo played a shoulder-hit at the fifth line that human commentators called a mistake. After 12 minutes of human deliberation, Lee Sedol returned to the board visibly shaken. Move 37 became shorthand for AI creativity — a play no human grandmaster would have chosen, and one that the network calculated had a one-in-ten-thousand chance of being played by a human.", "illustration": { "imageSearchTerm": "Go board stone placement", "imagePrompt": "A close-up of a Go board with black and white stones, focused on a single unusual stone placement mid-game.", "alt": "Move 37", "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Go_Nagai_20080704_Japan_Expo_01.jpg", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-move-37.webp" }, "studyGuideAnchor": "move-37", "uid": "x64nrk13cmgxa" }, { "id": "numeric-alphago-seoul", "shape": "numeric", "tags": [ "alphago", "milestones", "dates" ], "prompt": { "modality": "text", "value": "Year AlphaGo defeated Lee Sedol in Seoul" }, "value": 2016, "unit": "year", "uid": "17mz2b81crbc92" }, { "id": "mcq-alphago-nature", "shape": "mcq", "tags": [ "alphago", "milestones" ], "prompt": { "modality": "text", "value": "What journal made AlphaGo's first major paper a cover story in January 2016?" }, "options": [ { "modality": "text", "value": "PNAS" }, { "modality": "text", "value": "Nature" }, { "modality": "text", "value": "Cell" }, { "modality": "text", "value": "Science" } ], "correctIndex": 1, "uid": "74e1n2cwjcy0" }, { "id": "fact-openai-founding", "shape": "fact", "tags": [ "openai", "musk" ], "title": "OpenAI is born", "body": "In December 2015, Elon Musk and Sam Altman announced OpenAI: a non-profit AI lab pledging $1 billion in commitments, explicitly framed as a counterweight to Google DeepMind. Mallaby treats the move as a direct reaction to AlphaGo's progress — Musk in particular had decided that DeepMind \"under Google\" was the most dangerous concentration of AI talent in the world.", "illustration": { "imageSearchTerm": "Elon Musk Sam Altman", "imagePrompt": "Two entrepreneurs standing together at a press event announcing a new research initiative.", "alt": "OpenAI is born", "depictable": true, "credit": "Wikimedia Commons · CC BY 3.0", "creditUrl": "https://commons.wikimedia.org/wiki/File:Elon_Musk_2011_Shankbone.JPG", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-openai-founding.webp" }, "studyGuideAnchor": "openai-is-born", "uid": "1tog341oy0a92" }, { "id": "pair-altman", "shape": "pair", "tags": [ "openai", "altman", "people", "vocabulary" ], "sideA": { "modality": "text", "value": "OpenAI cofounder who became CEO" }, "sideB": { "modality": "text", "value": "Sam Altman" }, "uid": "crhgoaavi334" }, { "id": "pair-sutskever", "shape": "pair", "tags": [ "openai", "people", "vocabulary" ], "sideA": { "modality": "text", "value": "OpenAI's first chief scientist" }, "sideB": { "modality": "text", "value": "Ilya Sutskever" }, "uid": "287wt71dokfvd" }, { "id": "fact-poaching", "shape": "fact", "tags": [ "openai", "deepmind" ], "title": "The poaching war begins", "body": "OpenAI's first hires deliberately targeted the DeepMind alumni network. Ilya Sutskever, a Toronto-trained Hinton student who had been at Google, became chief scientist. Pieter Abbeel and others followed. DeepMind, under Google's pay scales, fought back by offering equity-style retention packages that were unusual for an arms-length subsidiary.", "illustration": { "imageSearchTerm": "Ilya Sutskever portrait", "imagePrompt": "A machine-learning researcher standing at a whiteboard covered in abstract sketches in a tech office.", "alt": "The poaching war begins", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/fact-poaching.jpg", "credit": "Pexels · Willians Huerta", "creditUrl": "https://www.pexels.com/photo/close-up-of-resting-rhino-in-sunlit-habitat-36531070/" }, "studyGuideAnchor": "the-poaching-war-begins", "uid": "1joybcb1az3fr5" }, { "id": "fact-musk-email", "shape": "fact", "tags": [ "openai", "musk", "milestones" ], "title": "September 2017: a fateful email", "body": "On 20 September 2017, Greg Brockman and Ilya Sutskever wrote to Musk and Altman. Their worry: a single individual could end up with \"absolute control\" of OpenAI. \"You are concerned that Demis could create an AGI dictatorship,\" Sutskever wrote of Musk's view, \"so do we.\" The implication: OpenAI itself needed protection from concentrated power.", "illustration": { "imageSearchTerm": "Greg Brockman portrait", "imagePrompt": "A tech executive typing intently on a laptop late at night in a dim office.", "alt": "September 2017: a fateful email", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Greg_Foran_and_Sanna_Marin_1.12.2022_-_52535562455.jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-musk-email.webp" }, "studyGuideAnchor": "september-2017-a-fateful-email", "uid": "ebad1193u9z3" }, { "id": "fact-restructure-2019", "shape": "fact", "tags": [ "openai", "altman", "governance" ], "title": "OpenAI's capped-profit pivot", "body": "In 2019, OpenAI restructured into a non-profit parent with a for-profit subsidiary capped at 100× return for early investors. Altman, now CEO, used the new structure to raise a $1 billion commitment from Microsoft. The move would later be cited as one reason the board lost confidence in him — they had been told the cap would hold.", "illustration": { "imageSearchTerm": "Sam Altman portrait", "imagePrompt": "A tech CEO in a casual button-down shirt speaking during a fireside chat on stage.", "alt": "OpenAI's capped-profit pivot", "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Sam_Waterson_1998.jpg", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-restructure-2019.webp" }, "studyGuideAnchor": "openai-s-capped-profit-pivot", "uid": "zz6ew9yuq2e" }, { "id": "fact-suleyman-departure", "shape": "fact", "tags": [ "suleyman", "deepmind", "governance" ], "title": "Suleyman exits, August 2019", "body": "On 21 August 2019, Bloomberg published a long story about bullying allegations against Mustafa Suleyman at DeepMind. By the end of the year he had moved to Google's Mountain View office in a vague vice-president role with no team. Eight years later he would emerge as CEO of Microsoft AI — the third corner of the original triangle, leading the partner of OpenAI.", "illustration": { "imageSearchTerm": "Google Mountain View campus", "imagePrompt": "A sprawling modern tech campus with colorful bicycles parked outside low-rise glass office buildings.", "alt": "Portrait of DeepMind co-founder Mustafa Suleyman", "credit": "Pexels · Bright and modern Google Store entrance with clear glass facade in Mountain View, California.", "creditUrl": "https://www.pexels.com/photo/google-store-entrance-at-mountain-view-california-32534024/", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-suleyman-departure.webp" }, "studyGuideAnchor": "suleyman-exits-august-2019", "uid": "owc7mp1gdozyf" }, { "id": "fact-google-merger", "shape": "fact", "tags": [ "google", "deepmind", "race" ], "title": "Google DeepMind, April 2023", "body": "After ChatGPT shocked Mountain View, Sundar Pichai consolidated. In April 2023 Google Brain — the team that had invented the transformer in 2017 — was merged into DeepMind to form a single unit called Google DeepMind. Hassabis became its CEO. The merger ended a decade-long internal rivalry and made him directly responsible for Google's response to OpenAI.", "illustration": { "imageSearchTerm": "Sundar Pichai portrait", "imagePrompt": "A tech company CEO speaking at a corporate keynote event on a well-lit stage.", "alt": "Google DeepMind, April 2023", "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Sundar_C_at_Muthina_Kathirikai_Audio_Launch.jpg", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-google-merger.webp" }, "studyGuideAnchor": "google-deepmind-april-2023", "uid": "kgu32a1rfxsg4" }, { "id": "fact-three-northlondoners", "shape": "fact", "tags": [ "deepmind", "openai", "people" ], "title": "Three north Londoners, three giants", "body": "Mallaby notes the parallel: by 2024 Hassabis ran Google DeepMind; Suleyman ran Microsoft AI; the third corner of the original DeepMind triangle, Legg, remained the lab's chief AGI scientist. \"Two north Londoners with immigrant parents,\" Mallaby writes, \"headed AI operations at two American tech giants.\"", "illustration": { "imageSearchTerm": "three paths diverging cities", "imagePrompt": "Three separate paths diverging from one starting point, each leading toward a different glowing city skyline in the distance.", "alt": "Three north Londoners, three giants", "depictable": false, "url": "https://cdn.recurxive.com/packs/infinity-machine/fact-three-northlondoners.jpg", "credit": "Pexels · Ray Zhu", "creditUrl": "https://www.pexels.com/photo/giant-outdoor-chess-game-in-london-mall-28798478/" }, "studyGuideAnchor": "three-north-londoners-three-giants", "uid": "x9dr8ls1ztsj" }, { "id": "fact-anthropic", "shape": "fact", "tags": [ "openai", "safety" ], "title": "Anthropic splits off, 2021", "body": "Dario Amodei and his sister Daniela led a group of safety-focused OpenAI researchers — including Paul Christiano and the future UK AI Safety Institute scientific chief Geoffrey Irving — out of OpenAI in 2021 to found Anthropic. Mallaby treats the spin-off as a recurring pattern: safety-minded technologists keep leaving the labs they helped build.", "illustration": { "imageSearchTerm": "Dario Amodei portrait", "imagePrompt": "A researcher in a button-down shirt speaking thoughtfully during a panel discussion.", "alt": "Anthropic splits off, 2021", "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Dario_Hysginon.jpg", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-anthropic.webp" }, "studyGuideAnchor": "anthropic-splits-off-2021", "uid": "197j1u31f0ft1p" }, { "id": "mcq-microsoft-ai-ceo", "shape": "mcq", "tags": [ "suleyman", "race" ], "prompt": { "modality": "text", "value": "Which DeepMind cofounder became CEO of Microsoft AI in 2024?" }, "options": [ { "modality": "text", "value": "Shane Legg" }, { "modality": "text", "value": "Mustafa Suleyman" }, { "modality": "text", "value": "David Silver" }, { "modality": "text", "value": "Demis Hassabis" } ], "correctIndex": 1, "uid": "xnusdh1drdy7j" }, { "id": "fact-altman-firing", "shape": "fact", "tags": [ "openai", "altman", "milestones" ], "title": "Five days that shook OpenAI", "body": "On 17 November 2023, OpenAI's board fired Sam Altman over a video call. Within forty-eight hours over 700 of OpenAI's 770 staff threatened to quit. Microsoft offered to absorb anyone who left. Five days later Altman was reinstated and the dissenting directors stepped down. Mallaby: \"The hopes of slowing the AI race had been destroyed.\"", "illustration": { "imageSearchTerm": "OpenAI office San Francisco", "imagePrompt": "A modern tech office building lobby with glass walls, seen from the street at dusk.", "alt": "Portrait of OpenAI CEO Sam Altman", "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:OpenAI_logo_with_magnifying_glass_(52916339167).jpg", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-altman-firing.webp" }, "factVariant": "image-heavy", "imageCaption": "Fired on a video call. 700 of 770 staff threatened to quit. Five days later he was back — and the brakes were gone.", "studyGuideAnchor": "five-days-that-shook-openai", "uid": "gwftrne8yfih" }, { "id": "numeric-altman-days", "shape": "numeric", "tags": [ "openai", "altman", "milestones" ], "prompt": { "modality": "text", "value": "Number of days between Altman's firing and reinstatement at OpenAI" }, "value": 5, "unit": "days", "uid": "1o9hlgbl2qlkp" }, { "id": "fact-alphafold-seoul", "shape": "fact", "tags": [ "alphafold", "milestones" ], "title": "A promise on a Seoul sidewalk", "body": "After AlphaGo defeated Lee Sedol in March 2016, Hassabis walked back to the hotel with David Silver. \"I'm telling you,\" he said, \"we can solve protein folding.\" Silver was sceptical. Hassabis had been waiting to tackle the problem since reading the literature as an undergraduate. Within months he had scrambled a hackathon team to try.", "illustration": { "imageSearchTerm": "Seoul city street night", "imagePrompt": "Two men walking together along a city sidewalk in Seoul at night, streetlights glowing around them.", "alt": "A promise on a Seoul sidewalk", "credit": "Pexels · A lively aerial view of a bustling street in Seoul at night, showcasing bright city lights and a vibrant atmosphere.", "creditUrl": "https://www.pexels.com/photo/street-in-seoul-south-korea-at-night-26063382/", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-alphafold-seoul.webp" }, "studyGuideAnchor": "a-promise-on-a-seoul-sidewalk", "uid": "3q0gvczicnzq" }, { "id": "def-protein-folding", "shape": "definition", "tags": [ "alphafold", "vocabulary" ], "term": { "modality": "text", "value": "Protein folding" }, "definition": { "modality": "text", "value": "The process by which a chain of amino acids twists itself into a specific three-dimensional shape that determines a protein's function. Predicting the final shape from the sequence alone — biology's Fermat's Last Theorem — was the challenge AlphaFold solved." }, "uid": "sbbs91hq951j" }, { "id": "fact-anfinsen", "shape": "fact", "tags": [ "alphafold" ], "title": "Anfinsen's conjecture", "body": "In his 1972 Nobel Prize lecture, Christian Anfinsen conjectured that a protein's sequence alone determined its folded shape — that the information needed was all encoded in the amino acids. For the next half-century, biologists believed him in principle but could not extract the shape from the sequence in practice. AlphaFold was the algorithmic vindication of Anfinsen.", "illustration": { "imageSearchTerm": "protein 3D structure model", "imagePrompt": "A colorful ribbon-diagram 3D model of a folded protein structure against a dark background.", "alt": "Anfinsen's conjecture", "credit": "Unsplash · National Cancer Institute · Unsplash License", "creditUrl": "https://unsplash.com/photos/purple-ribbon-eeEiys6TU3c", "depictable": true, "subject": "Computer-rendered ribbon diagram of a single folded protein (purple helices and strands) with a small ball-and-stick ligand bound in its central pocket, teal background", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-anfinsen.webp" }, "studyGuideAnchor": "anfinsen-s-conjecture", "uid": "1cmwu483w230y" }, { "id": "pair-jumper", "shape": "pair", "tags": [ "alphafold", "jumper", "people", "vocabulary" ], "sideA": { "modality": "text", "value": "Lead scientist on AlphaFold 2" }, "sideB": { "modality": "text", "value": "John Jumper" }, "uid": "c0ti8pdacpyz" }, { "id": "fact-jumper-arrives", "shape": "fact", "tags": [ "alphafold", "jumper", "milestones" ], "title": "Jumper arrives, October 2017", "body": "Andrew Senior, the project's first lead, recruited John Jumper in October 2017. Jumper, then 32, had worked across mathematics, physics, chemistry, biology and machine learning — including three years at D. E. Shaw Research, where he had tried to model proteins with Newtonian physics. He brought biological intuition the team had lacked.", "illustration": { "imagePrompt": "A scientist in a lab coat studying a 3D molecular structure displayed on a computer monitor.", "imageSearchTerm": "John Jumper portrait", "alt": "A 3D ribbon diagram of a folded protein", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:John_Pierce_St._John_(1833-1916)_(10506774896).jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-jumper-arrives.webp" }, "factVariant": "image-heavy", "imageCaption": "Andrew Senior, the project's first lead, recruited John Jumper in October 2017.", "studyGuideAnchor": "jumper-arrives-october-2017", "uid": "tbsdw1xmelyb" }, { "id": "fact-casp", "shape": "fact", "tags": [ "alphafold", "milestones" ], "title": "CASP: the biennial test", "body": "Every two years since 1994, Professor John Moult had organised CASP — the Critical Assessment of Structure Prediction. Competing labs received amino acid sequences for proteins whose true shapes were known only to the experimentalists. Predictions were scored by Global Distance Test (GDT). For AlphaFold, CASP was the externally verifiable scoreboard.", "illustration": { "imageSearchTerm": "scientists analyzing protein structure", "imagePrompt": "Researchers gathered around computer monitors displaying colorful 3D protein structure models, comparing predictions.", "alt": "CASP: the biennial test", "credit": "Pexels · Two scientists working together in a laboratory, analyzing data and discussing results.", "creditUrl": "https://www.pexels.com/photo/medical-practitioners-working-together-8442378/", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-casp.webp" }, "studyGuideAnchor": "casp-the-biennial-test", "uid": "1545gmh1l9wnl7" }, { "id": "def-gdt", "shape": "definition", "tags": [ "alphafold", "vocabulary" ], "term": { "modality": "text", "value": "GDT (Global Distance Test)" }, "definition": { "modality": "text", "value": "CASP's headline accuracy score. It measures the percentage of a predicted protein's main-chain atoms that lie within a tolerated distance of the ground-truth crystallography structure. Roughly 90 GDT is the threshold at which a prediction is as good as an experimental measurement." }, "uid": "1o1jzpcd15nk6" }, { "id": "fact-distogram", "shape": "fact", "tags": [ "alphafold" ], "title": "From contact map to distogram", "body": "Where rival labs trained networks to predict whether two amino acids touched, AlphaFold trained its convolutional network to predict the distance between them on a continuous scale. \"Like going from black and white to a full-colour TV,\" researcher Marek Barwinski said. The shift to distograms was AlphaFold 1's signature insight.", "illustration": { "imageSearchTerm": "vintage black white television", "imagePrompt": "A television screen transitioning from black-and-white static on one half to vivid full color on the other half.", "alt": "From contact map to distogram", "credit": "Pexels · Black and white image of an old television with antenna on a stand.", "creditUrl": "https://www.pexels.com/photo/gray-scale-photo-analogue-of-television-3151392/", "depictable": false, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-distogram.webp" }, "studyGuideAnchor": "from-contact-map-to-distogram", "uid": "nojo15yvexjn" }, { "id": "fact-cancun-2018", "shape": "fact", "tags": [ "alphafold", "milestones" ], "title": "AlphaFold wins CASP13, December 2018", "body": "In Cancun in early December 2018, AlphaFold bested the other 97 CASP entrants — particularly in the \"free modelling\" category where no evolutionary template was known. DeepMind won 25 of 43 hard cases; its nearest rival won three. \"What just happened?\" became the question of the conference.", "illustration": { "imageSearchTerm": "science conference presentation hall", "imagePrompt": "A packed conference hall where scientists are presenting research results on a large screen to an audience of peers.", "alt": "AlphaFold wins CASP13, December 2018", "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:About_Time_Tour_Gottenheim_03.jpg", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-cancun-2018.webp" }, "studyGuideAnchor": "alphafold-wins-casp13-december-2018", "uid": "1ctfsqe8mhg3w" }, { "id": "fact-direct-folding", "shape": "fact", "tags": [ "alphafold" ], "title": "The direct-folding pivot", "body": "After CASP13, Hassabis told the team they were not done. The next iteration would predict the exact 3D coordinates of every atom directly, instead of relying on a post-hoc search. The GDT score initially crashed from 60 to 20. The team kept going. By late 2019 the new approach was back at 60 and climbing.", "illustration": { "imageSearchTerm": "graph line falling rising", "imagePrompt": "An abstract glowing line curving sharply downward then climbing steadily back up and beyond its starting point.", "alt": "The direct-folding pivot", "credit": "Unsplash · Austin Hervias · Unsplash License", "creditUrl": "https://unsplash.com/photos/candlestick-stock-chart-on-dark-screen-VLpWpv3oDB4", "depictable": false, "subject": "A dark-screen candlestick chart with two faint horizontal dotted lines: candles slide down from top-left into a near-vertical red crash at one-third across, then climb steadily back up to the right edge; no legible text, dates, ticker or watermark", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-direct-folding.webp" }, "studyGuideAnchor": "the-direct-folding-pivot", "uid": "1rrp0tes2nvlo" }, { "id": "fact-tetraformer", "shape": "fact", "tags": [ "alphafold", "transformer" ], "title": "Enter the transformer", "body": "In 2019 Jumper's team rebuilt AlphaFold's neural network as a transformer — specifically a \"tetraformer\" combining four variants of the architecture. They masked amino acids in the UniProt database and trained the model to guess what was missing, the same trick BERT had used on language. Evolutionary patterns turned out to encode the geometry of proteins.", "illustration": { "imageSearchTerm": "abstract neural network structure", "imagePrompt": "An abstract web of glowing interconnected nodes and attention lines forming a layered network structure, pulsing with data flow.", "alt": "Enter the transformer", "depictable": false, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:FEQ_July_2018_Brockhampton_(44828738891).jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-tetraformer.webp" }, "studyGuideAnchor": "enter-the-transformer", "uid": "144pis11r67qgr" }, { "id": "fact-casp14", "shape": "fact", "tags": [ "alphafold", "milestones" ], "title": "CASP14, November 2020", "body": "In November 2020, Professor Moult tallied the CASP14 scores. AlphaFold 2 had scored 92.4 — more than 50 percent higher than any previous best in the contest's history. The accuracy was good enough that an experimentalist's gold-standard test became unable to detect a meaningful difference. \"A seismic and unprecedented shift,\" one observer wrote.", "illustration": { "imageSearchTerm": "AlphaFold protein structure model", "imagePrompt": "A colorful 3D ribbon-diagram visualization of a folded protein structure displayed on a computer screen.", "alt": "CASP14, November 2020", "credit": "Pexels · Maksim Goncharenok", "creditUrl": "https://www.pexels.com/photo/november-inscription-made-from-leaves-5552810/", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/fact-casp14.jpg" }, "factVariant": "image-heavy", "imageCaption": "92.4 — more than 50% above any score in the contest's history. \"A seismic and unprecedented shift.\"", "studyGuideAnchor": "casp14-november-2020", "uid": "fhcf491hwf1qf" }, { "id": "numeric-casp14-score", "shape": "numeric", "tags": [ "alphafold", "milestones" ], "prompt": { "modality": "text", "value": "AlphaFold 2's GDT score at CASP14" }, "value": 92, "unit": "GDT", "tolerance": 1, "uid": "nfgioe1garcpo" }, { "id": "fact-database", "shape": "fact", "tags": [ "alphafold", "milestones" ], "title": "200 million proteins", "body": "After CASP14, DeepMind partnered with the European Bioinformatics Institute to release the AlphaFold structure database. Within months it held all 20,000 human proteins; by July 2022 it had folded around 200 million — almost every protein known to science. Over 3 million researchers had used it by 2025.", "illustration": { "imageSearchTerm": "protein structure database visualization", "imagePrompt": "A gallery grid of colorful 3D protein structure models representing a vast digital database.", "alt": "A protein structure predicted by AlphaFold", "depictable": true, "credit": "Wikimedia Commons · CC BY-SA 4.0", "creditUrl": "https://commons.wikimedia.org/wiki/File:CHOP_protein_structure.png", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-database.webp" }, "factVariant": "image-heavy", "imageCaption": "200 million structures by July 2022 — nearly every protein known to science, free to 3 million researchers.", "studyGuideAnchor": "200-million-proteins", "uid": "vpg2nm1rqf2yw" }, { "id": "numeric-proteins", "shape": "numeric", "tags": [ "alphafold", "milestones" ], "prompt": { "modality": "text", "value": "Approximate number of proteins folded by AlphaFold by July 2022 (millions)" }, "value": 200, "unit": "million", "tolerance": 20, "uid": "he81he17vlsck" }, { "id": "fact-cheese-tomorrow", "shape": "fact", "tags": [ "alphafold", "hassabis" ], "title": "\"Cheese tomorrow\"", "body": "Asked late in the book about language models, Hassabis was dismissive: \"Everyone talks about the benefits of language models, but mostly it's just cheese tomorrow. The clearest benefit from AI so far is AlphaFold.\" He wants to spend the next decade compressing a century of medical discovery — and pointing the technology at diseases of the developing world.", "illustration": { "imageSearchTerm": "medical research laboratory scientist", "imagePrompt": "A scientist in a research laboratory studying a molecular structure on a computer screen, representing accelerated medical discovery.", "alt": "\"Cheese tomorrow\"", "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Kaufmann_Medical_Building.jpg", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-cheese-tomorrow.webp" }, "studyGuideAnchor": "cheese-tomorrow", "uid": "re5p0pqakg0f" }, { "id": "fact-chatgpt-shock", "shape": "fact", "tags": [ "openai", "race", "milestones" ], "title": "The ChatGPT shock", "body": "On 30 November 2022, OpenAI released ChatGPT — initially as a research preview of a polished GPT-3.5. Within five days it had a million users; within two months, 100 million. Mallaby treats the launch as the moment the AI race went public: governments, journalists and venture capitalists who had ignored AI suddenly demanded daily updates.", "illustration": { "imageSearchTerm": "ChatGPT app interface screen", "imagePrompt": "A smartphone screen displaying a minimalist chat interface with rows of message bubbles, text left illegibly blurred.", "alt": "The ChatGPT shock", "credit": "Pexels · Smartphone with ChatGPT app interface next to eyeglasses on patterned surface.", "creditUrl": "https://www.pexels.com/photo/smartphone-with-chat-gpt-ai-system-16094062/", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-chatgpt-shock.webp" }, "factVariant": "image-heavy", "imageCaption": "1 million users in 5 days. 100 million in 2 months. The day the AI race went public.", "studyGuideAnchor": "the-chatgpt-shock", "uid": "1qszwqjh6mg9t" }, { "id": "pair-irving", "shape": "pair", "tags": [ "safety", "people", "vocabulary" ], "sideA": { "modality": "text", "value": "DeepMind safety pioneer who joined the UK AI Safety Institute" }, "sideB": { "modality": "text", "value": "Geoffrey Irving" }, "uid": "9yg7ue1mov88c" }, { "id": "fact-irving-arrives", "shape": "fact", "tags": [ "safety", "deepmind" ], "title": "Geoffrey Irving joins DeepMind", "body": "In autumn 2019 DeepMind hired Geoffrey Irving from OpenAI, where he had been part of a safety circle with Dario Amodei and Paul Christiano. Irving's goal: design alignment into the architecture itself. \"If a malevolent actor solves the problem first,\" he believed, \"civilisation would be in trouble.\"", "illustration": { "imageSearchTerm": "Geoffrey Irving DeepMind researcher", "imagePrompt": "A portrait of a computer scientist in a modern AI research office, thoughtful expression, softly lit.", "alt": "Geoffrey Irving joins DeepMind", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-irving-arrives.webp", "credit": "Pexels · File:Geoffrey Arend 2011.jpg", "creditUrl": "https://commons.wikimedia.org/wiki/File:Geoffrey_Arend_2011.jpg" }, "studyGuideAnchor": "geoffrey-irving-joins-deepmind", "uid": "12ong9d127r74j" }, { "id": "def-alignment", "shape": "definition", "tags": [ "safety", "vocabulary" ], "term": { "modality": "text", "value": "Alignment" }, "definition": { "modality": "text", "value": "The technical and conceptual problem of ensuring that an AI system reliably pursues human-intended goals, even as the system becomes more capable. Irving's analogy: rules of safety might be simple (do not harm, do not deceive), but implementing them is as hard as AlphaGo's mastery of Go." }, "uid": "14xrds51o881l7" }, { "id": "fact-one-sentence", "shape": "fact", "tags": [ "safety", "milestones" ], "title": "The one-sentence letter", "body": "In May 2023 the Center for AI Safety released a one-sentence letter: \"Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war.\" Hinton, Bengio, Hassabis, Altman, Amodei and hundreds of others signed. Mallaby treats it as the rhetorical inflection point that forced governments into the conversation.", "illustration": { "imageSearchTerm": "AI extinction risk statement", "imagePrompt": "Many silhouetted hands raised together against a dark sky, symbolizing a shared warning voiced by many people.", "alt": "The one-sentence letter", "depictable": false, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-one-sentence.webp", "credit": "Pexels · File:Edmund Blair Leighton - A summer shower.jpg", "creditUrl": "https://commons.wikimedia.org/wiki/File:Edmund_Blair_Leighton_-_A_summer_shower.jpg" }, "studyGuideAnchor": "the-one-sentence-letter", "uid": "1li9f5n1j5bug5" }, { "id": "fact-buchanan", "shape": "fact", "tags": [ "buchanan", "governance", "milestones" ], "title": "An AI czar in the White House", "body": "In June 2023 the Biden administration created a new position for an AI czar and filled it with Ben Buchanan, a Georgetown professor of AI and cybersecurity, who had been working at the National Security Council. Buchanan's mandate was to address the labs' Hinton–Bengio concerns using the leverage of government — without strangling the technology.", "illustration": { "imagePrompt": "The White House building in Washington DC, iconic columned facade, viewed at dusk.", "imageSearchTerm": "White House Washington DC", "alt": "The White House illuminated at dusk", "depictable": true, "credit": "Pexels · Front view of the iconic White House in Washington, DC, showing its lawn and fountain.", "creditUrl": "https://www.pexels.com/photo/the-white-house-6477549/", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-buchanan.webp" }, "factVariant": "image-heavy", "imageCaption": "In 2023 the White House got its first AI czar: a Georgetown professor named Ben Buchanan.", "studyGuideAnchor": "an-ai-czar-in-the-white-house", "uid": "1pt7dpmv6kj0o" }, { "id": "fact-voluntary-commitments", "shape": "fact", "tags": [ "buchanan", "governance", "milestones" ], "title": "Voluntary commitments, 21 July 2023", "body": "On 21 July 2023, leaders from Google, OpenAI, Anthropic, Meta, Amazon, Microsoft, and Suleyman's Inflection signed voluntary commitments at the White House: pre-release safety testing, cybersecurity investment, public reporting of capabilities. Not a single major US frontier lab declined. The pledges were thin on enforcement but, Buchanan reasoned, public backsliding would now be visible.", "illustration": { "imageSearchTerm": "White House tech leaders meeting", "imagePrompt": "A formal meeting room at a government building with technology executives seated around a long table.", "alt": "Voluntary commitments, 21 July 2023", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-voluntary-commitments.webp", "credit": "Pexels · File:MaskedLynxAvatar.png", "creditUrl": "https://commons.wikimedia.org/wiki/File:MaskedLynxAvatar.png" }, "studyGuideAnchor": "voluntary-commitments-21-july-2023", "uid": "1h7ornj1xa0nph" }, { "id": "fact-executive-order", "shape": "fact", "tags": [ "buchanan", "governance", "milestones" ], "title": "Biden's AI executive order, October 2023", "body": "In October 2023 Buchanan orchestrated an executive order invoking the Defense Production Act. Developers training the most powerful AI models would have to notify the government and share red-team results. The order also created what would become the US AI Safety Institute. The administration was building the rails before Congress could be convinced to legislate.", "illustration": { "imageSearchTerm": "president signing executive order", "imagePrompt": "A close-up of a hand signing an official document at a desk in the Oval Office, the paper's text left softly out of focus.", "alt": "Official portrait of President Joe Biden", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/fact-executive-order.jpg", "credit": "Wikimedia Commons (Wikipedia: Joe Biden)", "creditUrl": "https://en.wikipedia.org/wiki/Joe_Biden" }, "studyGuideAnchor": "biden-s-ai-executive-order-october-2023", "uid": "t1683g1cj8yqa" }, { "id": "fact-bletchley", "shape": "fact", "tags": [ "bletchley", "governance", "milestones" ], "title": "Bletchley Park summit, November 2023", "body": "On 1 November 2023, delegates from 28 countries gathered at Bletchley Park — Alan Turing's wartime workplace — for the first international AI Safety Summit, conceived six months earlier by Hassabis in a conversation with Prime Minister Rishi Sunak. Yoshua Bengio jabbed his finger at DeepMind safety advisers; safety institutes in the US and UK were announced. The Bletchley Declaration was signed.", "illustration": { "imagePrompt": "The historic redbrick Victorian mansion of Bletchley Park in England, surrounded by manicured lawns.", "imageSearchTerm": "Bletchley Park mansion England", "alt": "The Victorian manor house at Bletchley Park in autumn", "depictable": true, "credit": "Pexels · Elegant view of Cragside House, a historic mansion in Rothbury, England.", "creditUrl": "https://www.pexels.com/photo/cragside-house-in-countryside-in-uk-20528754/", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-bletchley.webp" }, "factVariant": "image-heavy", "imageCaption": "28 nations gather at Turing's wartime codebreaking hall for the first global summit on AI safety.", "studyGuideAnchor": "bletchley-park-summit-november-2023", "uid": "1lnaqa8115kfam" }, { "id": "numeric-bletchley-countries", "shape": "numeric", "tags": [ "bletchley", "governance", "milestones" ], "prompt": { "modality": "text", "value": "Number of countries that signed the 2023 Bletchley Declaration" }, "value": 28, "unit": "countries", "uid": "1801ykscs87vm" }, { "id": "fact-china-policy", "shape": "fact", "tags": [ "race", "governance" ], "title": "Chip exports and China", "body": "In October 2022, a month before ChatGPT, the Biden administration rolled out a wide-ranging ban on the supply of advanced semiconductors to China. Mallaby reads the move as the closing of the most consequential lane of US-China cooperation. By the time Hassabis pitched Sunak on a global AI summit, the geopolitics had already hardened.", "illustration": { "imageSearchTerm": "semiconductor microchip close-up", "imagePrompt": "A macro close-up of an advanced semiconductor computer chip mounted on a circuit board.", "alt": "Chip exports and China", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Embedded_World_KNX_TP_Demo_Board.jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-china-policy.webp" }, "studyGuideAnchor": "chip-exports-and-china", "uid": "tdsw4v9yoai1" }, { "id": "fact-sputnik", "shape": "fact", "tags": [ "race", "alphago" ], "title": "China's Sputnik moment", "body": "Hassabis has been aware of China's determination to compete since 2017. AlphaGo's defeat of Ke Jie that year was widely described inside Beijing as a Sputnik moment. By 2022 the State Council had a national AI strategy with 2030 ambitions. The race the West thought it was running was, from Beijing, a race China had already been preparing for.", "illustration": { "imageSearchTerm": "Go board tournament match", "imagePrompt": "A professional Go player intensely studying a wooden Go board covered with black and white stones.", "alt": "China's Sputnik moment", "credit": "Pexels · Margo Evardson", "creditUrl": "https://www.pexels.com/photo/candid-street-conversation-in-shanghai-37337981/", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/fact-sputnik.jpg" }, "studyGuideAnchor": "china-s-sputnik-moment", "uid": "1mtdztkzgj39m" }, { "id": "mcq-ai-safety-institute", "shape": "mcq", "tags": [ "bletchley", "safety", "governance" ], "prompt": { "modality": "text", "value": "Which two countries announced national AI Safety Institutes around the November 2023 Bletchley summit?" }, "options": [ { "modality": "text", "value": "United Kingdom and United States" }, { "modality": "text", "value": "Israel and Singapore" }, { "modality": "text", "value": "Canada and Japan" }, { "modality": "text", "value": "France and Germany" } ], "correctIndex": 0, "uid": "1v88u0t1l110d7" }, { "id": "fact-gemini-launch", "shape": "fact", "tags": [ "gemini", "milestones" ], "title": "Gemini, December 2023", "body": "On 6 December 2023 Google DeepMind launched Gemini, its first frontier model designed jointly by the merged London and Mountain View teams. Two smaller versions shipped immediately; the headline Gemini Ultra was previewed against GPT-4 on MMLU. The launch was Hassabis's response to ChatGPT — late, but the first proof that the merger could ship at all.", "illustration": { "imageSearchTerm": "Google Gemini AI launch", "imagePrompt": "A dark stage at a technology product launch event, illuminated by dramatic blue and purple light beams.", "alt": "The Google Gemini chatbot interface", "depictable": true, "credit": "Wikipedia — Google Gemini · See Wikimedia Commons", "creditUrl": "https://en.wikipedia.org/wiki/Google_Gemini", "subject": "Screenshot of the Google Gemini web chatbot: Gemini logo and wordmark, sidebar (New chat, Search chats, Library, Notebooks), 'Where should we start?' heading, 'Ask Gemini' input with a 'Flash' model picker and mic, 'Upgrade' button; a user avatar labelled 'Amanda' bottom-left", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-gemini-launch.webp" }, "studyGuideAnchor": "gemini-december-2023", "uid": "1nj94p91x0sy13" }, { "id": "def-mmlu", "shape": "definition", "tags": [ "gemini", "vocabulary" ], "term": { "modality": "text", "value": "MMLU (Massive Multitask Language Understanding)" }, "definition": { "modality": "text", "value": "A benchmark of 57 subjects spanning maths, humanities, ethics and law. In 2020, GPT-3 scored 44%; in 2023 GPT-4 reached 86%; Gemini Ultra hit 90%, the first AI system to exceed top human experts. Each generational bump on MMLU is roughly a year of student-level cognition gained." }, "uid": "1m1msnbhnpkal" }, { "id": "numeric-gemini-mmlu", "shape": "numeric", "tags": [ "gemini", "milestones" ], "prompt": { "modality": "text", "value": "Gemini Ultra's announced MMLU score (percentage)" }, "value": 90, "unit": "%", "tolerance": 1, "uid": "uy5rxjfqcadh" }, { "id": "fact-merger-culture", "shape": "fact", "tags": [ "gemini", "deepmind" ], "title": "Cultures collide", "body": "Building Gemini was as much a management challenge as a research one. Mountain View's culture was \"crash projects out quickly\"; London's was \"measure each tweak\". Mallaby describes 80-hour weeks, two-time-zone Slack channels, and Sergey Brin re-engaging \"after years of playboy life\" to clear roadblocks. Speed eventually won the internal debates.", "illustration": { "imageSearchTerm": "engineers working late office", "imagePrompt": "A busy open-plan tech office of engineers working at their desks late into the night.", "alt": "Cultures collide", "credit": "Pexels · An architect working late at night in an office, using a laptop and surrounded by building models.", "creditUrl": "https://www.pexels.com/photo/a-man-working-in-the-office-6615044/", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-merger-culture.webp" }, "studyGuideAnchor": "cultures-collide", "uid": "1etakhoum01z6" }, { "id": "fact-silver-quits-rlhf", "shape": "fact", "tags": [ "gemini", "deepmind" ], "title": "David Silver and the RL wall", "body": "After ChatGPT, David Silver — the AlphaGo veteran — agreed to lead Gemini's post-training. He hoped to apply sophisticated machine RL to language models. After six months of frustration, he quit the post-training team. \"It wasn't clear how machine RL could distinguish a good haiku from a great one,\" Mallaby explains. A Bard veterans team that focused on cleaning data and RLHF widgets shipped the upgrade instead.", "illustration": { "imageSearchTerm": "AI researcher conference presentation", "imagePrompt": "A computer scientist presenting research on stage at a technology conference, a large screen behind them.", "alt": "David Silver and the RL wall", "credit": "Pexels · Anat Landa", "creditUrl": "https://www.pexels.com/photo/entrance-to-the-tower-of-david-museum-jerusalem-38430952/", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/fact-silver-quits-rlhf.jpg" }, "studyGuideAnchor": "david-silver-and-the-rl-wall", "uid": "k1vfvx4tw2fb" }, { "id": "fact-mech-interp", "shape": "fact", "tags": [ "safety", "deepmind" ], "title": "Mechanistic interpretability", "body": "Geoffrey Irving's safety group at DeepMind extended a line of research started at Anthropic: identifying the specific circuits inside a network responsible for specific behaviours. Mallaby calls the field — mechanistic interpretability — \"chinks of light in the black box\". By 2024 it was one of the few alignment programmes with credible commercial uptake.", "illustration": { "imageSearchTerm": "neural network circuit visualization", "imagePrompt": "An abstract, colorful visualization of interconnected neural network nodes and glowing pathways, like a circuit diagram.", "alt": "Mechanistic interpretability", "credit": "Pexels · Elegant 3D visualization of neural networks showcasing abstract connections in a digital space.", "creditUrl": "https://www.pexels.com/photo/an-artist-s-illustration-of-artificial-intelligence-ai-this-image-was-inspired-neural-networks-used-in-deep-learning-it-was-created-by-novoto-studio-as-part-of-the-visualising-ai-proje-17483871/", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-mech-interp.webp" }, "studyGuideAnchor": "mechanistic-interpretability", "uid": "1mtfxlk6cvldy" }, { "id": "fact-gemini-15", "shape": "fact", "tags": [ "gemini", "milestones" ], "title": "Gemini 1.5 Pro, February 2024", "body": "On 15 February 2024 Google DeepMind released Gemini 1.5 Pro — a million-token context window, sparse mixture-of-experts architecture, and the first model from a London-led, scientifically minded \"strike team\" of the kind that had built AlphaFold. Among the team were Sebastian Borgeaud and Jack Rae, who had returned from a stint at OpenAI.", "illustration": { "imageSearchTerm": "Gemini 1.5 Pro launch", "imagePrompt": "A futuristic screen displaying flowing abstract streams of glowing data, representing a vast digital information window.", "alt": "Gemini 1.5 Pro, February 2024", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:AstronautAssignmentsChart.PNG", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-gemini-15.webp" }, "studyGuideAnchor": "gemini-1-5-pro-february-2024", "uid": "15ue3attsk4xj" }, { "id": "fact-still-first-innings", "shape": "fact", "tags": [ "gemini", "race", "hassabis" ], "title": "\"First innings of the game\"", "body": "Mallaby's Hassabis, on Christmas vacation in late 2023: \"OpenAI had the scaling and engineering and product focus that we didn't have. But now with Gemini we are matching that. And I think we still have the better ideas.\" He insisted ChatGPT was \"just the start\" — the AI race was in its first innings, not its ninth.", "illustration": { "imageSearchTerm": "tech CEO interview candid", "imagePrompt": "A confident tech executive speaking candidly during an interview, seated in a modern office.", "alt": "\"First innings of the game\"", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/fact-still-first-innings.jpg", "credit": "Pexels · Lesandu Alokabandara", "creditUrl": "https://www.pexels.com/photo/photo-of-athletes-walking-on-the-field-during-a-cricket-match-21415109/" }, "studyGuideAnchor": "first-innings-of-the-game", "uid": "1agxnfv1047m85" }, { "id": "fact-altman-untrustworthy", "shape": "fact", "tags": [ "openai", "altman" ], "title": "Why the OpenAI board fired Altman", "body": "Mallaby identifies two triggers. First, undisclosed fundraising overtures to investors in the Persian Gulf. Second, a pattern of trying to push out independent board members who criticised him. The board, charged with \"safe AI for the benefit of the world\", concluded Altman could not be trusted with the mission. They underestimated how thoroughly Silicon Valley would close ranks behind him.", "illustration": { "imageSearchTerm": "tech CEO press conference", "imagePrompt": "A tech executive speaking at a press conference behind a podium, serious and composed expression.", "alt": "Why the OpenAI board fired Altman", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-altman-untrustworthy.webp", "credit": "Pexels · File:Id tech 6 (beta) logo.png", "creditUrl": "https://commons.wikimedia.org/wiki/File:Id_tech_6_(beta)_logo.png" }, "studyGuideAnchor": "why-the-openai-board-fired-altman", "uid": "fou1yrm6r31d" }, { "id": "fact-nadella-offer", "shape": "fact", "tags": [ "openai", "altman", "milestones" ], "title": "Nadella's lifeline", "body": "Within 36 hours of Altman's firing, Satya Nadella publicly invited him to lead a new Microsoft AI lab with a budget to hire \"anyone from OpenAI who wanted to join him.\" Three days later more than 700 of OpenAI's 770 employees had signed a letter threatening to follow. The board folded; Altman returned.", "illustration": { "imageSearchTerm": "tech CEO conference stage", "imagePrompt": "A tech executive speaking confidently on stage at a corporate conference, dark suit, warm stage lighting.", "alt": "Portrait of Microsoft CEO Satya Nadella", "credit": "Pexels · Speaker presenting at a tech conference, engaging with the audience.", "creditUrl": "https://www.pexels.com/photo/conference-speaker-at-technology-event-34774349/", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-nadella-offer.webp" }, "studyGuideAnchor": "nadella-s-lifeline", "uid": "1ypnooti7rttb" }, { "id": "fact-conclusion", "shape": "fact", "tags": [ "intro", "race", "hassabis" ], "title": "Mallaby's closing register", "body": "Mallaby ends the book in 2025–26, with Hassabis simultaneously a Nobel laureate and the chief executive of one of the most powerful AI labs in the world. The book closes on a paradox he does not pretend to resolve: that the most morally serious major figure in AI is also racing as hard as anybody to build the technology he has spent his life warning the world about.", "illustration": { "imageSearchTerm": "Nobel Prize ceremony medal", "imagePrompt": "A scientist proudly holding a gold medal at a formal awards ceremony, elegant hall in the background.", "alt": "Mallaby's closing register", "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Female_Nobel_literature_prize_laureates_by_decade.jpg", "depictable": true, "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fact-conclusion.webp" }, "studyGuideAnchor": "mallaby-s-closing-register", "uid": "1dpcza11lewxnr" }, { "id": "pair-hassabis-q", "shape": "pair", "tags": [ "deepmind", "hassabis", "vocabulary" ], "sideA": { "modality": "text", "value": "DeepMind cofounder who won the 2024 Nobel Prize in Chemistry" }, "sideB": { "modality": "text", "value": "Demis Hassabis" }, "uid": "11vlcnsa9og7a" }, { "id": "pair-legg-q", "shape": "pair", "tags": [ "deepmind", "legg", "vocabulary" ], "sideA": { "modality": "text", "value": "Cofounder who coined \"AGI\"" }, "sideB": { "modality": "text", "value": "Shane Legg" }, "uid": "j2ab717o2jrd" }, { "id": "pair-suleyman-q", "shape": "pair", "tags": [ "deepmind", "suleyman", "vocabulary" ], "sideA": { "modality": "text", "value": "Cofounder who became Microsoft AI CEO" }, "sideB": { "modality": "text", "value": "Mustafa Suleyman" }, "uid": "1dbcyyze795ql" }, { "id": "pair-silver-q", "shape": "pair", "tags": [ "deepmind", "alphago", "vocabulary" ], "sideA": { "modality": "text", "value": "Cambridge friend who led AlphaGo" }, "sideB": { "modality": "text", "value": "David Silver" }, "uid": "wh36nhprg1q3" }, { "id": "pair-jumper-q", "shape": "pair", "tags": [ "deepmind", "alphafold", "vocabulary" ], "sideA": { "modality": "text", "value": "Scientist who led AlphaFold 2" }, "sideB": { "modality": "text", "value": "John Jumper" }, "uid": "x8d6jr1rdavlh" }, { "id": "pair-thiel-q", "shape": "pair", "tags": [ "policy", "thiel", "vocabulary" ], "sideA": { "modality": "text", "value": "First investor in DeepMind" }, "sideB": { "modality": "text", "value": "Peter Thiel" }, "uid": "1umpcbi12ujt20" }, { "id": "pair-mnih-q", "shape": "pair", "tags": [ "deepmind", "dqn", "vocabulary" ], "sideA": { "modality": "text", "value": "Researcher who built the Atari DQN" }, "sideB": { "modality": "text", "value": "Vlad Mnih" }, "uid": "9h5k5rdjniix" }, { "id": "pair-altman-q", "shape": "pair", "tags": [ "openai", "altman", "vocabulary" ], "sideA": { "modality": "text", "value": "OpenAI CEO fired and rehired in 2023" }, "sideB": { "modality": "text", "value": "Sam Altman" }, "uid": "1de1p5ggmjjce" }, { "id": "pair-sutskever-q", "shape": "pair", "tags": [ "openai", "vocabulary" ], "sideA": { "modality": "text", "value": "OpenAI cofounder who led the board revolt" }, "sideB": { "modality": "text", "value": "Ilya Sutskever" }, "uid": "shxkvpgox90n" }, { "id": "pair-buchanan-q", "shape": "pair", "tags": [ "policy", "buchanan", "vocabulary" ], "sideA": { "modality": "text", "value": "Biden's AI czar from June 2023" }, "sideB": { "modality": "text", "value": "Ben Buchanan" }, "uid": "6oeslh15z8i0r" }, { "id": "pair-irving-q", "shape": "pair", "tags": [ "deepmind", "safety", "vocabulary" ], "sideA": { "modality": "text", "value": "DeepMind safety lead who joined UK AISI" }, "sideB": { "modality": "text", "value": "Geoffrey Irving" }, "uid": "f7sx0cy7vnem" }, { "id": "pair-bletchley-q", "shape": "pair", "tags": [ "policy", "bletchley", "vocabulary" ], "sideA": { "modality": "text", "value": "Site of the first global AI Safety Summit" }, "sideB": { "modality": "text", "value": "Bletchley Park" }, "uid": "1igxvi010px2o2" }, { "id": "num-hassabis-born", "shape": "numeric", "tags": [ "dates", "people" ], "prompt": { "modality": "text", "value": "Year Demis Hassabis was born" }, "value": 1976, "unit": "year", "tolerance": 3, "uid": "10r3e2ex4lqsg" }, { "id": "num-suleyman-born", "shape": "numeric", "tags": [ "dates", "people" ], "prompt": { "modality": "text", "value": "Year Mustafa Suleyman was born" }, "value": 1984, "unit": "year", "tolerance": 3, "uid": "acd6q6iucifw" }, { "id": "num-kasparov-beaten", "shape": "numeric", "tags": [ "dates", "milestones" ], "prompt": { "modality": "text", "value": "Year Deep Blue defeated Garry Kasparov" }, "value": 1997, "unit": "year", "tolerance": 3, "uid": "161a2ehxah2d7" }, { "id": "num-deepmind-founded", "shape": "numeric", "tags": [ "dates", "milestones" ], "prompt": { "modality": "text", "value": "Year DeepMind was founded" }, "value": 2010, "unit": "year", "tolerance": 2, "uid": "1r7pvdcn91nui" }, { "id": "num-alexnet", "shape": "numeric", "tags": [ "dates", "milestones" ], "prompt": { "modality": "text", "value": "Year AlexNet won the ImageNet competition" }, "value": 2012, "unit": "year", "tolerance": 2, "uid": "zypuq7asghzx" }, { "id": "num-google-acquires", "shape": "numeric", "tags": [ "dates", "milestones" ], "prompt": { "modality": "text", "value": "Year Google acquired DeepMind" }, "value": 2014, "unit": "year", "tolerance": 2, "uid": "1q7c13zcu886t" }, { "id": "num-dqn-nature", "shape": "numeric", "tags": [ "dates", "milestones" ], "prompt": { "modality": "text", "value": "Year DQN was published in Nature" }, "value": 2015, "unit": "year", "tolerance": 2, "uid": "4iprxvtr4av" }, { "id": "num-alphago-seoul", "shape": "numeric", "tags": [ "dates", "milestones" ], "prompt": { "modality": "text", "value": "Year AlphaGo beat Lee Sedol in Seoul" }, "value": 2016, "unit": "year", "tolerance": 2, "uid": "1ttwv4q1u79f54" }, { "id": "num-transformer-paper", "shape": "numeric", "tags": [ "dates", "milestones" ], "prompt": { "modality": "text", "value": "Year the Transformer paper was published" }, "value": 2017, "unit": "year", "tolerance": 2, "uid": "ev16j411hdl2y" }, { "id": "num-chatgpt-launch", "shape": "numeric", "tags": [ "dates", "milestones" ], "prompt": { "modality": "text", "value": "Year ChatGPT was released" }, "value": 2022, "unit": "year", "tolerance": 2, "uid": "1m3ecn419yixeq" }, { "id": "num-google-paid", "shape": "numeric", "tags": [ "money", "milestones" ], "prompt": { "modality": "text", "value": "How many millions of US dollars Google paid for DeepMind" }, "value": 600, "unit": "million USD", "tolerance": 100, "uid": "ne0hnd1lnoqsf" }, { "id": "num-dqn-games", "shape": "numeric", "tags": [ "scale", "milestones" ], "prompt": { "modality": "text", "value": "Number of Atari 2600 games the single DQN architecture was trained and evaluated on" }, "value": 49, "unit": "games", "tolerance": 5, "uid": "1fapmp56rqrwf" }, { "id": "num-dqn-human-level", "shape": "numeric", "tags": [ "scale", "milestones" ], "prompt": { "modality": "text", "value": "Number of Atari games on which DQN reached human-level or better" }, "value": 29, "unit": "games", "tolerance": 5, "uid": "njy25979039r" }, { "id": "cz-geb-book", "shape": "cloze", "tags": [ "influences", "book" ], "template": "Mallaby singles out Douglas Hofstadter's ___ as the single most important book in Hassabis's intellectual formation.", "answer": "Gödel, Escher, Bach", "distractors": [ "The Mind's I", "Society of Mind", "On Intelligence" ], "explanation": "GEB connected logic, art, and music into a single argument about how mind emerges from pattern.", "uid": "1jz28ll1nx74q3" }, { "id": "cz-elixir-year", "shape": "cloze", "tags": [ "hassabis", "career" ], "template": "Hassabis founded Elixir Studios in ___ at age 22.", "answer": "1998", "distractors": [ "1996", "2000", "2002" ], "explanation": "Elixir launched on 7 July 1998 in London with funding from Eidos.", "uid": "gkr51vgbfu0h" }, { "id": "cz-deepmind-google", "shape": "cloze", "tags": [ "deepmind", "acquisition" ], "template": "Google acquired DeepMind in ___ for a reported $600 million.", "answer": "2014", "distractors": [ "2012", "2013", "2016" ], "explanation": "The deal closed in January 2014, with Larry Page personally championing the acquisition.", "uid": "b9a41oefny22" }, { "id": "cz-alphago-sedol", "shape": "cloze", "tags": [ "alphago", "milestones" ], "template": "AlphaGo defeated Lee Sedol in Seoul in ___ , a moment Mallaby treats as the public arrival of deep RL.", "answer": "2016", "distractors": [ "2015", "2017", "2018" ], "explanation": "AlphaGo took the five-game match 4–1 in March 2016.", "uid": "19tgkdy1sd7esc" }, { "id": "cz-alphafold-casp", "shape": "cloze", "tags": [ "alphafold", "milestones" ], "template": "AlphaFold 2 stunned biologists at the ___ protein-structure competition in 2020 with a GDT score around 92.", "answer": "CASP14", "distractors": [ "CASP12", "ImageNet", "MLPerf" ], "explanation": "CASP is the biennial Critical Assessment of Structure Prediction; CASP14 was the breakout.", "uid": "243hygny9u42" }, { "id": "cz-bletchley-declaration", "shape": "cloze", "tags": [ "governments", "safety" ], "template": "Twenty-eight countries signed the ___ Declaration on AI safety in November 2023.", "answer": "Bletchley", "distractors": [ "Geneva", "Brussels", "London" ], "explanation": "The summit was held at Bletchley Park, the wartime codebreaking site.", "uid": "r09x8naqx6cp" }, { "id": "concept-demis-hassabis", "shape": "concept", "tags": [ "key-concept", "person" ], "conceptKind": "person", "name": "Demis Hassabis", "clues": [ "He is a co-founder of a leading AI company.", "He received the 2024 Nobel Prize in Chemistry for his work on AlphaFold.", "He was a chess prodigy, mastering the game by age 13." ], "uid": "1suvnc71urgwah" }, { "id": "concept-shane-legg", "shape": "concept", "tags": [ "key-concept", "person" ], "conceptKind": "person", "name": "Shane Legg", "clues": [ "He is known for coining a term related to advanced AI.", "He co-founded a prominent AI research lab in London.", "He proposed the concept of Artificial General Intelligence during discussions in 2002." ], "uid": "13vi5gvhpqqad" }, { "id": "concept-mustafa-suleyman", "shape": "concept", "tags": [ "key-concept", "person" ], "conceptKind": "person", "name": "Mustafa Suleyman", "clues": [ "He is one of the co-founders of a well-known AI company.", "He has a background in conflict resolution and diplomacy.", "He was instrumental in pitching to a billionaire investor for initial funding." ], "uid": "8piorhscgkrr" }, { "id": "concept-david-silver", "shape": "concept", "tags": [ "key-concept", "person" ], "conceptKind": "person", "name": "David Silver", "clues": [ "He is a key figure in the development of AI at a major lab.", "He led a project that achieved a significant milestone in the game Go.", "He met the main subject of the book during their university years." ], "uid": "1m52prx1sftcnb" }, { "id": "concept-vlad-mnih", "shape": "concept", "tags": [ "key-concept", "person" ], "conceptKind": "person", "name": "Vlad Mnih", "clues": [ "He is recognized for his pioneering work in reinforcement learning.", "He led a team that created a neural network to play classic video games.", "His research contributed to a major breakthrough in AI gaming." ], "uid": "ixb5pp10ejvrz" }, { "id": "concept-john-jumper", "shape": "concept", "tags": [ "key-concept", "person" ], "conceptKind": "person", "name": "John Jumper", "clues": [ "Recruited to a London AI lab in 2017, he led its leap into structural biology.", "He had bounced across physics and chemistry before finding his defining problem.", "In 2024 he shared the Nobel Prize in Chemistry with the lab's founder for predicting protein shapes." ], "uid": "1snz6im17znll0" }, { "id": "concept-google-buys-deepmind", "shape": "concept", "tags": [ "key-concept", "event" ], "conceptKind": "event", "name": "Google buys DeepMind", "clues": [ "In January 2014, a major tech company made a significant acquisition in the AI field.", "This acquisition was Europe's largest at the time, valued at around £400 million.", "The deal included a condition for an independent ethics board to oversee AI applications." ], "uid": "1c3sxy31ij9n7t" }, { "id": "concept-the-day-pong-clicked", "shape": "concept", "tags": [ "key-concept", "event" ], "conceptKind": "event", "name": "The day Pong clicked", "clues": [ "On 7 July 2013, a significant milestone in AI development occurred during a major tennis event.", "A neural network learned to play a classic arcade game from scratch, impressing its creators.", "By the evening, the AI was able to beat the built-in opponent, moving towards human-level performance." ], "uid": "yk2mh01c8r4um" }, { "id": "concept-singularity-summit", "shape": "concept", "tags": [ "key-concept", "event" ], "conceptKind": "event", "name": "Singularity Summit", "clues": [ "In August 2010, a significant conference focused on the future of technology took place in San Francisco.", "This event featured discussions on advanced AI and its implications for society.", "A trio of AI researchers pitched a groundbreaking lab concept aimed at building AGI safely." ], "uid": "13m0luas2cmt0" }, { "id": "concept-the-halloween-scenario", "shape": "concept", "tags": [ "key-concept", "event" ], "conceptKind": "event", "name": "The Halloween Scenario", "clues": [ "In October 2009, a lecture at UCL addressed the urgent need for AI safety planning.", "The presenter argued that advanced AI was no longer just a theoretical concept.", "This talk sparked ongoing discussions about the implications of powerful AI technologies." ], "uid": "6u8eg51wesu13" }, { "id": "concept-the-atari-strike-team", "shape": "concept", "tags": [ "key-concept", "event" ], "conceptKind": "event", "name": "The Atari strike team", "clues": [ "In 2013, a small team at a leading AI lab tackled a challenge involving classic video games.", "They aimed to create a single neural network that could learn to play multiple games.", "This project marked the lab's first major public success in reinforcement learning." ], "uid": "pqed2ydthlc" }, { "id": "concept-musk-vs-page-on-the-lawn", "shape": "concept", "tags": [ "key-concept", "event" ], "conceptKind": "event", "name": "Musk vs Page on the lawn", "clues": [ "In October 2012, a dinner conversation at a tech mogul's home shifted to the topic of AI.", "One guest warned about the potential dangers of advanced AI, while the host had a contrasting view.", "This discussion is seen as the moment a key alliance in AI development began to fracture." ], "uid": "1mv094b1q7teyl" }, { "id": "def-deepmind", "shape": "definition", "tags": [ "deepmind", "hassabis" ], "term": { "modality": "text", "value": "DeepMind" }, "definition": { "modality": "text", "value": "DeepMind is the London AI lab Demis Hassabis co-founded in 2010 and Google acquired in 2014. DeepMind's stated mission is to solve intelligence and then use it to solve everything else." }, "curatedDistractors": [ "OpenAI", "Anthropic", "Cohere" ], "uid": "vs41dj1f64qnd" }, { "id": "def-alphago", "shape": "definition", "tags": [ "alphago", "milestones" ], "term": { "modality": "text", "value": "AlphaGo" }, "definition": { "modality": "text", "value": "AlphaGo is the DeepMind program that defeated world champion Lee Sedol at the board game Go in 2016. AlphaGo combined deep neural networks with Monte Carlo tree search." }, "curatedDistractors": [ "AlphaFold", "Deep Blue", "Watson" ], "uid": "lu56ucgsjtke" }, { "id": "def-alphafold", "shape": "definition", "tags": [ "alphafold", "milestones" ], "term": { "modality": "text", "value": "AlphaFold" }, "definition": { "modality": "text", "value": "AlphaFold is the DeepMind system that predicts a protein's three-dimensional structure from its amino-acid sequence. AlphaFold cracked a fifty-year grand challenge in biology and helped win Hassabis a share of the 2024 Nobel Prize in Chemistry." }, "curatedDistractors": [ "AlphaGo", "AlphaZero", "Watson" ], "uid": "3dtd7ude1tqk" }, { "id": "def-reinforcement-learning", "shape": "definition", "tags": [ "atari", "dqn" ], "term": { "modality": "text", "value": "Reinforcement learning" }, "definition": { "modality": "text", "value": "Reinforcement learning is a training method in which an agent learns by trial and error, receiving rewards for good actions and penalties for bad ones. DeepMind used reinforcement learning to master Atari games directly from raw pixels." }, "curatedDistractors": [ "Supervised learning", "Unsupervised learning", "Transfer learning" ], "uid": "htdcq01e1egji" }, { "id": "def-gemini", "shape": "definition", "tags": [ "gemini", "google" ], "term": { "modality": "text", "value": "Gemini" }, "definition": { "modality": "text", "value": "Gemini is Google DeepMind's flagship family of multimodal AI models, launched in 2023. Gemini was the lab's answer to OpenAI's GPT-4 and marked its return to the research frontier." }, "curatedDistractors": [ "Claude", "LLaMA", "PaLM" ], "uid": "7bzaz18h6sgj" }, { "id": "def-elixir-studios", "shape": "definition", "tags": [ "elixir", "biography" ], "term": { "modality": "text", "value": "Elixir Studios" }, "definition": { "modality": "text", "value": "Elixir Studios was the games company Hassabis founded after leaving Bullfrog. Elixir Studios built ambitious AI-driven games but struggled commercially and closed in 2005." }, "curatedDistractors": [ "Lionhead Studios", "Rockstar Games", "Rare" ], "uid": "9p17ysphygty" }, { "id": "def-casp", "shape": "definition", "tags": [ "alphafold", "milestones" ], "term": { "modality": "text", "value": "CASP" }, "definition": { "modality": "text", "value": "CASP is the biennial Critical Assessment of Structure Prediction contest, where protein-folding methods compete on unsolved structures. AlphaFold's runaway victory at CASP in 2020 stunned the field of biology." }, "curatedDistractors": [ "ImageNet Challenge", "Netflix Prize", "DARPA Grand Challenge" ], "uid": "na0qunp2oc9t" }, { "id": "scrapbook-infinity-machine-1", "shape": "fact", "factVariant": "scrapbook", "title": "The Sweetness of Discovery", "body": "Exploring the allure of knowledge and breakthroughs in AI.", "tags": [ "scrapbook", "overview" ], "illustration": { "kind": "scrapbook", "alt": "Exploring the allure of knowledge and breakthroughs in AI.", "imagePrompt": "A cheerful scrapbook collage featuring a hand-lettered title 'The Sweetness of Discovery' as the focal point. The background includes vibrant illustrations of key discoveries like chess, Atari consoles, and neural networks. Labels in a playful font highlight milestones like 'AlphaGo' and 'AlphaFold'. Doodles of light bulbs and stars add a creative touch, while torn-paper layers in warm colors bring depth. Arrows guide the eye through the visuals, showcasing the exciting journey of AI advancements.", "imageSearchTerm": "infinity-machine scrapbook poster", "credit": "AI-generated (gpt-image-1)", "url": "https://cdn.recurxive.com/packs/infinity-machine/generated/scrapbook-1.webp" }, "uid": "1mx7rvm157iioc" }, { "id": "scrapbook-infinity-machine-2", "shape": "fact", "factVariant": "scrapbook", "title": "Born in Finchley, July 1976", "body": "The early beginnings of Demis Hassabis's journey.", "tags": [ "scrapbook", "overview" ], "illustration": { "kind": "scrapbook", "alt": "The early beginnings of Demis Hassabis's journey.", "imagePrompt": "A hand-crafted scrapbook poster capturing the childhood of Demis Hassabis. The title, 'Born in Finchley, July 1976', is expressed in bold, hand-lettered style. The collage includes charming illustrations of Finchley, a young boy playing chess, and a vintage ZX Spectrum. Short, handwritten labels denote 'Chess Prodigy' and 'Self-Taught Coder'. A warm palette of yellows and browns creates a nostalgic feel. The design incorporates fun doodles of chess pieces and coding symbols to emphasize the early influence of games and technology.", "imageSearchTerm": "infinity-machine scrapbook poster", "credit": "AI-generated (gpt-image-1)", "url": "https://cdn.recurxive.com/packs/infinity-machine/generated/scrapbook-2.webp" }, "uid": "okyzmfjs5ywx" }, { "id": "q-im-deepmind-firstname", "shape": "pair", "sideA": { "modality": "text", "value": "DeepMind's first working name" }, "sideB": { "modality": "text", "value": "Solaria, taken from the Asimov novel", "short": "Solaria" }, "tags": [ "deepmind", "vocabulary" ], "uid": "x0kz0xiyaidn" }, { "id": "q-im-deepmind-registered", "shape": "numeric", "prompt": { "modality": "text", "value": "Year the DeepMind name was registered" }, "value": 2010, "unit": "year", "tags": [ "deepmind", "vocabulary" ], "tolerance": 0, "uid": "b19wkz15bsj31" }, { "id": "q-im-musk-page-dinner", "shape": "pair", "sideA": { "modality": "text", "value": "Where Musk warned AI could become a digital god" }, "sideB": { "modality": "text", "value": "At a dinner at Larry Page's home in October 2012", "short": "Page's home, 2012" }, "tags": [ "google", "musk" ], "uid": "ja877j1tma2f5" }, { "id": "q-im-brockman-email", "shape": "pair", "sideA": { "modality": "text", "value": "Who wrote the September 2017 email to Musk and Altman" }, "sideB": { "modality": "text", "value": "Greg Brockman and Ilya Sutskever", "short": "Brockman and Sutskever" }, "tags": [ "openai", "musk", "milestones" ], "uid": "1vz9t0fey1cgp" }, { "id": "q-im-openai-cap", "shape": "pair", "sideA": { "modality": "text", "value": "OpenAI's 2019 investor return cap" }, "sideB": { "modality": "text", "value": "Profit capped at 100 times for early investors", "short": "100x cap" }, "tags": [ "openai", "altman", "governance" ], "uid": "1d0ftqp1y7efqb" }, { "id": "q-im-openai-restructure-year", "shape": "numeric", "prompt": { "modality": "text", "value": "Year OpenAI restructured into a capped-profit company" }, "value": 2019, "unit": "year", "tags": [ "openai", "altman", "governance" ], "tolerance": 0, "uid": "1s7gfgr109avb1" }, { "id": "q-im-google-deepmind-merger", "shape": "pair", "sideA": { "modality": "text", "value": "April 2023 merger that formed Google DeepMind" }, "sideB": { "modality": "text", "value": "Google Brain merged into DeepMind under Hassabis", "short": "Brain into DeepMind" }, "tags": [ "google", "deepmind", "race" ], "uid": "p460uq1minl78" }, { "id": "q-im-buchanan-czar", "shape": "pair", "sideA": { "modality": "text", "value": "Biden's June 2023 AI czar" }, "sideB": { "modality": "text", "value": "Ben Buchanan, a Georgetown AI and cybersecurity professor", "short": "Ben Buchanan" }, "tags": [ "buchanan", "governance", "milestones" ], "uid": "1v99ouw1ih7q6i" }, { "id": "q-im-voluntary-commitments-date", "shape": "pair", "sideA": { "modality": "text", "value": "Where AI labs signed voluntary commitments in July 2023" }, "sideB": { "modality": "text", "value": "At the White House, with no major US frontier lab declining", "short": "The White House" }, "tags": [ "buchanan", "governance", "milestones" ], "uid": "1cfh60m1v0wk1s" }, { "id": "q-im-bletchley-year", "shape": "numeric", "prompt": { "modality": "text", "value": "Year of the first international AI Safety Summit at Bletchley Park" }, "value": 2023, "unit": "year", "tags": [ "bletchley", "governance", "milestones" ], "tolerance": 0, "uid": "1oncgxm19bfqbc" }, { "id": "q-im-chip-ban-year", "shape": "numeric", "prompt": { "modality": "text", "value": "Year the Biden administration banned advanced semiconductor supply to China" }, "value": 2022, "unit": "year", "tags": [ "race", "governance" ], "tolerance": 0, "uid": "2wd8nwk43sla" }, { "id": "q-im-gemini15-release", "shape": "pair", "sideA": { "modality": "text", "value": "Gemini 1.5 Pro's headline capability at launch" }, "sideB": { "modality": "text", "value": "A million-token context window with a mixture-of-experts design", "short": "Million-token context" }, "tags": [ "gemini", "milestones" ], "uid": "6rwfcecr2o94" }, { "id": "q-im-gemini15-year", "shape": "numeric", "prompt": { "modality": "text", "value": "Year Google DeepMind released Gemini 1.5 Pro" }, "value": 2024, "unit": "year", "tags": [ "gemini", "milestones" ], "tolerance": 0, "uid": "aq61jb13jd9s9" }, { "id": "q-im-altman-firing-triggers", "shape": "pair", "sideA": { "modality": "text", "value": "Two triggers Mallaby gives for the Altman firing" }, "sideB": { "modality": "text", "value": "Undisclosed Gulf fundraising and pushing out critical board members", "short": "Gulf funding and board purge" }, "tags": [ "openai", "altman" ], "uid": "2v8xxerkhw0g" }, { "id": "q-im-hassabis-nobel", "shape": "pair", "sideA": { "modality": "text", "value": "Hassabis's dual status as the book closes in 2025-26" }, "sideB": { "modality": "text", "value": "A Nobel laureate and chief executive of a leading AI lab", "short": "Nobel laureate and CEO" }, "tags": [ "intro", "race", "hassabis" ], "uid": "1bktusq1kq4bec" }, { "id": "concept-infinity-machine-reinforcement-learning", "shape": "concept", "tags": [ "key-concept", "thing" ], "conceptKind": "thing", "name": "Reinforcement Learning", "clues": [ "A type of machine learning where agents learn by interacting with an environment.", "It focuses on taking suitable actions to maximize cumulative reward.", "This learning paradigm was crucial for DeepMind's early successes." ], "uid": "1lgtjso9kcibq" }, { "id": "concept-infinity-machine-deep-learning", "shape": "concept", "tags": [ "key-concept", "thing" ], "conceptKind": "thing", "name": "Deep Learning", "clues": [ "A subset of machine learning inspired by the structure of the human brain.", "It uses neural networks with many layers to analyze various forms of data.", "This technique underpins many of DeepMind's advancements, including AlphaGo." ], "uid": "11gqq1kbjje9q" }, { "id": "concept-infinity-machine-neural-network", "shape": "concept", "tags": [ "key-concept", "thing" ], "conceptKind": "thing", "name": "Neural Network", "clues": [ "A computational model whose design is inspired by the way the human brain processes information.", "They consist of interconnected nodes or 'neurons' that work together to solve complex problems.", "DeepMind utilized this structure to develop systems like AlphaFold." ], "uid": "oh81js7zaw9y" }, { "id": "concept-infinity-machine-alphago", "shape": "concept", "tags": [ "key-concept", "thing" ], "conceptKind": "thing", "name": "AlphaGo", "clues": [ "An AI program developed by DeepMind to play the game of Go.", "It famously defeated the world champion Lee Sedol in 2016, marking a significant milestone in AI.", "This program showcased the power of reinforcement learning and deep learning combined." ], "uid": "f9819d1kdcdov" }, { "id": "concept-infinity-machine-dqn-deep-q-network", "shape": "concept", "tags": [ "key-concept", "thing" ], "conceptKind": "thing", "name": "DQN (Deep Q-Network)", "clues": [ "A reinforcement learning algorithm that combines Q-learning with deep learning techniques.", "It was designed to learn how to play Atari games directly from raw pixel data.", "This architecture was pivotal in demonstrating the potential of general AI systems." ], "uid": "15bng5q1ofdd60" }, { "id": "concept-infinity-machine-alphafold", "shape": "concept", "tags": [ "key-concept", "thing" ], "conceptKind": "thing", "name": "AlphaFold", "clues": [ "An AI system developed by DeepMind to predict protein structures.", "It utilizes deep learning to solve complex biological problems.", "This technology has significant implications for drug discovery and biology." ], "uid": "hbhhml98c64z" }, { "id": "concept-infinity-machine-agi-artificial-general-intelligence", "shape": "concept", "tags": [ "key-concept", "thing" ], "conceptKind": "thing", "name": "AGI (Artificial General Intelligence)", "clues": [ "A theoretical form of AI that can understand, learn, and apply intelligence across a wide range of tasks.", "It aims to replicate human cognitive abilities in machines.", "DeepMind's mission is often associated with the pursuit of this concept." ], "uid": "rjsgf7ffxf3h" }, { "id": "mcq-rw-l2-p1-finchley", "shape": "mcq", "tags": [ "hassabis", "chess", "cognition" ], "prompt": { "modality": "text", "value": "Beyond calculation skills, what less obvious insight did Mallaby say competitive chess gave the young Hassabis?" }, "options": [ { "modality": "text", "value": "A preference for teamwork over solitary calculation", "short": "Teamwork beats solo calculation" }, { "modality": "text", "value": "A belief that computers could never rival human strategic intuition", "short": "Computers can't rival intuition" }, { "modality": "text", "value": "An intuition that human cognition has ceilings — that the bottleneck was hardware, not effort", "short": "Minds have hardware-set ceilings" }, { "modality": "text", "value": "A conviction that raw memorization beats intuition in complex games", "short": "Memorization beats intuition" } ], "correctIndex": 2, "explanation": "Playing chess, Hassabis could feel his own brain straining at certain positions in a way that suggested the limit was hardware, not effort — an early hint that intelligence might be something you could build rather than a fixed gift.", "uid": "n2da91ht637f" }, { "id": "mcq-rw-l3-p1-cambridge", "shape": "mcq", "tags": [ "hassabis", "neuroscience", "deepmind" ], "prompt": { "modality": "text", "value": "Hassabis's UCL neuroscience PhD — patients with hippocampal damage couldn't imagine future scenes — led him to see the brain as what, an idea that shaped DeepMind's agenda?" }, "options": [ { "modality": "text", "value": "A relational database of stored memories" }, { "modality": "text", "value": "A blank slate shaped entirely by experience" }, { "modality": "text", "value": "A generative model that constructs memories" }, { "modality": "text", "value": "A rule-based expert system" } ], "correctIndex": 2, "explanation": "Finding that the hippocampus both stores memories and constructs hypothetical futures convinced Hassabis the brain is a generative model — it builds scenes rather than retrieving them — a conviction that quietly structured DeepMind's research.", "uid": "7mrgi4bz49xy" }, { "id": "mcq-rw-l4-p1-legg", "shape": "mcq", "tags": [ "legg", "agi", "terminology" ], "prompt": { "modality": "text", "value": "Around 2002, in what context did Shane Legg propose the phrase 'Artificial General Intelligence'?" }, "options": [ { "modality": "text", "value": "Christening a recurrent-neural-network project with Jürgen Schmidhuber", "short": "An RNN project with Schmidhuber" }, { "modality": "text", "value": "Naming a new AI research lab he was founding", "short": "Naming an AI lab he founded" }, { "modality": "text", "value": "Titling his doctoral dissertation at IDSIA", "short": "Titling his PhD thesis (IDSIA)" }, { "modality": "text", "value": "Brainstorming a title for a planned book of essays with Marcus Hutter", "short": "Essay-book title with Hutter" } ], "correctIndex": 3, "explanation": "While Legg and Marcus Hutter were brainstorming a title for a planned essay collection around 2002, Legg proposed 'Artificial General Intelligence.' The phrase stuck and became the field's name for its most ambitious goal.", "uid": "ncxdbgbr835e" }, { "id": "mcq-rw-l5-p1-thiel", "shape": "mcq", "tags": [ "deepmind", "thiel", "funding" ], "prompt": { "modality": "text", "value": "Which investor's fund led DeepMind's earliest outside funding?" }, "options": [ { "modality": "text", "value": "Sequoia Capital" }, { "modality": "text", "value": "Marc Andreessen's Andreessen Horowitz" }, { "modality": "text", "value": "Elon Musk's personal investment" }, { "modality": "text", "value": "Peter Thiel's Founders Fund" } ], "correctIndex": 3, "explanation": "Peter Thiel's Founders Fund led DeepMind's earliest outside funding — a notable irony given Thiel's later public skepticism about unchecked AI. Google acquired the lab in January 2014 for roughly £400 million.", "uid": "1czqnapd6dcnn" }, { "id": "mcq-rw-l5-p2-google", "shape": "mcq", "tags": [ "dqn", "experience-replay", "reinforcement-learning" ], "prompt": { "modality": "text", "value": "Why was experience replay key to making Deep Q-Networks learn stably?" }, "options": [ { "modality": "text", "value": "It let the network memorize game-specific rules for each Atari title", "short": "Memorized rules for each Atari game" }, { "modality": "text", "value": "It slowed the learning rate so the network never overshot", "short": "Slowed learning rate to stop overshoot" }, { "modality": "text", "value": "It stored past experiences in a buffer and trained on random samples, breaking the correlation between consecutive frames", "short": "Random-sampled buffer breaks correlation" }, { "modality": "text", "value": "It ran many agents in parallel and averaged their scores", "short": "Ran parallel agents, averaged scores" } ], "correctIndex": 2, "explanation": "Networks trained on sequential frames overfit to recent play and forget earlier lessons. Storing experiences and sampling random batches decorrelated the data — closer to independent, identically distributed examples — making deep reinforcement learning reliable.", "uid": "1hn39ns1ntzgke" }, { "id": "mcq-rw-l6-p1-atari", "shape": "mcq", "tags": [ "alphago", "move-37", "go" ], "prompt": { "modality": "text", "value": "AlphaGo's Move 37 against Lee Sedol was so unusual that its policy network estimated a human would play it about how often?" }, "options": [ { "modality": "text", "value": "Once in 100 times" }, { "modality": "text", "value": "Once in 10,000 times" }, { "modality": "text", "value": "Once in 1,000 times" }, { "modality": "text", "value": "Once in a million times" } ], "correctIndex": 1, "explanation": "Move 37 — a shoulder-hit on the fifth line — was called a mistake by human commentators, but AlphaGo's policy network put the odds a human would choose it at roughly one in ten thousand. It won territory human intuition would not reach.", "uid": "1itnqkm1or9brg" }, { "id": "mcq-rw-l7-p1-musk", "shape": "mcq", "tags": [ "openai", "musk", "split" ], "prompt": { "modality": "text", "value": "According to the section, what did Musk do after failing to gain operational control of OpenAI in early 2018?" }, "options": [ { "modality": "text", "value": "Founded xAI as a direct competitor" }, { "modality": "text", "value": "Sued Altman for breach of the nonprofit mission" }, { "modality": "text", "value": "Doubled his financial commitment to keep influence" }, { "modality": "text", "value": "Withdrew his promised funding and resigned from the board" } ], "correctIndex": 3, "explanation": "Musk pushed to run OpenAI himself; when he couldn't win operational control in early 2018, he withdrew his promised funding and resigned from the board — the rupture that quietly reshaped the industry's power map.", "uid": "k4s701k5zodu" }, { "id": "mcq-rw-l8-p1-promise", "shape": "mcq", "tags": [ "alphafold", "protein-folding", "distogram" ], "prompt": { "modality": "text", "value": "AlphaFold 1 broke from rival labs by having its network predict what about pairs of amino acids?" }, "options": [ { "modality": "text", "value": "Which amino acid would mutate next over evolution", "short": "Which amino acid mutates next" }, { "modality": "text", "value": "A binary yes-or-no of whether the two were in contact", "short": "Whether the two are in contact" }, { "modality": "text", "value": "The exact 3D coordinates of each atom directly", "short": "Each atom's 3D coordinates" }, { "modality": "text", "value": "The continuous distance between them, producing a \"distogram\"", "short": "Their distance (a distogram)" } ], "correctIndex": 3, "explanation": "Rivals predicted a binary contact map (in contact or not). AlphaFold instead predicted the distance between amino-acid pairs on a continuous scale — a distogram — described as going from black-and-white to full-colour TV. It was AlphaFold 1's defining architectural choice.", "uid": "yq9o5j1xa5dxh" }, { "id": "mcq-rw-l8-p2-casp14", "shape": "mcq", "tags": [ "alphafold", "casp", "gdt" ], "prompt": { "modality": "text", "value": "When DeepMind pivoted AlphaFold 2 to predict every atom's 3D coordinates directly, what happened to its GDT score at first?" }, "options": [ { "modality": "text", "value": "It rose smoothly with no setback" }, { "modality": "text", "value": "It jumped straight from 60 to 92" }, { "modality": "text", "value": "It held steady near 60 for two years" }, { "modality": "text", "value": "It crashed from about 60 to about 20 before clawing back" } ], "correctIndex": 3, "explanation": "The direct-folding pivot initially tanked the GDT score from ~60 to ~20. The team pressed on; by late 2019 it had recovered to 60 and kept climbing — a textbook case of accepting short-term regression to unlock a better approach.", "uid": "1qiouar0aw4c" }, { "id": "mcq-rw-l9-p1-shock", "shape": "mcq", "tags": [ "chatgpt", "adoption", "openai" ], "prompt": { "modality": "text", "value": "How fast did ChatGPT reach 100 million users after its 30 November 2022 launch?" }, "options": [ { "modality": "text", "value": "Within a year" }, { "modality": "text", "value": "Within five days" }, { "modality": "text", "value": "Within two months" }, { "modality": "text", "value": "Within one week" } ], "correctIndex": 2, "explanation": "ChatGPT hit 1 million users in five days and 100 million within two months — the fastest consumer-technology adoption on record, which is what pulled investors and governments off the sidelines and into the AI race.", "uid": "tt7leqyj8f3g" }, { "id": "mcq-rw-l10-p1-gemini", "shape": "mcq", "tags": [ "gemini", "reinforcement-learning", "david-silver" ], "prompt": { "modality": "text", "value": "According to Mallaby, why did David Silver's reinforcement-learning approach fail to crack Gemini's post-training?" }, "options": [ { "modality": "text", "value": "Reinforcement learning could not scale to a million-token context", "short": "RL couldn't scale to 1M tokens" }, { "modality": "text", "value": "Language quality offers no clear reward signal for RL to optimize", "short": "Language quality has no clear reward" }, { "modality": "text", "value": "Human raters generated too much data to process", "short": "Raters made too much data to process" }, { "modality": "text", "value": "The pretrained base model was never shared with his team", "short": "Base model wasn't shared with team" } ], "correctIndex": 1, "explanation": "RL needs a clear reward signal, but language quality resists scoring — as Mallaby puts it, RL couldn't tell a good haiku from a great one. Silver quit after ~6 months; a Bard team using data cleaning and RLHF shipped the upgrade instead.", "uid": "7m7797zegal1" }, { "id": "mcq-rw-l10-p2-end", "shape": "mcq", "tags": [ "hassabis", "mallaby", "ai-stance" ], "prompt": { "modality": "text", "value": "In the book's epilogue, which label does Hassabis say he prefers for his own stance on AI?" }, "options": [ { "modality": "text", "value": "Optimist" }, { "modality": "text", "value": "Doomer" }, { "modality": "text", "value": "Accelerationist" }, { "modality": "text", "value": "Realist" } ], "correctIndex": 3, "explanation": "Mallaby closes with Hassabis rejecting both \"doomer\" and \"accelerationist,\" preferring \"realist\" — fitting for a man who is at once a Nobel laureate warning about AI and a CEO racing to build it.", "uid": "3m0gme1srur7c" }, { "id": "mcq-rw-l11-p1-recall", "shape": "mcq", "tags": [ "deepmind", "google", "acquisition" ], "prompt": { "modality": "text", "value": "Roughly how much did Google pay to bring DeepMind in-house in 2014?" }, "options": [ { "modality": "text", "value": "About £1.2 billion" }, { "modality": "text", "value": "About £400 million" }, { "modality": "text", "value": "About £40 million" }, { "modality": "text", "value": "About £4 billion" } ], "correctIndex": 1, "explanation": "Larry Page and Sergey Brin acquired DeepMind for roughly £400 million in 2014 — Europe's largest AI acquisition at the time, made on the condition of an independent ethics board.", "uid": "30chwubs6cnc" }, { "id": "mcq-rw-lc-people", "shape": "mcq", "tags": [ "suleyman", "deepmind", "cofounders" ], "prompt": { "modality": "text", "value": "Which DeepMind cofounder came from a background in conflict resolution and diplomacy?" }, "options": [ { "modality": "text", "value": "Shane Legg" }, { "modality": "text", "value": "Mustafa Suleyman" }, { "modality": "text", "value": "Vlad Mnih" }, { "modality": "text", "value": "David Silver" } ], "correctIndex": 1, "explanation": "Mustafa Suleyman, the third cofounder and the lab's operator, had a background in conflict resolution and diplomacy before helping start DeepMind; he was instrumental in pitching the early billionaire investor and later ran Microsoft AI.", "uid": "15stmlx5mutcz" }, { "id": "mcq-rw-lc-things", "shape": "mcq", "tags": [ "dqn", "reinforcement-learning", "atari" ], "prompt": { "modality": "text", "value": "What made DeepMind's DQN (Deep Q-Network) notable in how it learned to play Atari games?" }, "options": [ { "modality": "text", "value": "It was trained on recordings of expert human players" }, { "modality": "text", "value": "It was given the rules of each game in advance" }, { "modality": "text", "value": "It read the game's internal RAM state" }, { "modality": "text", "value": "It learned directly from raw screen pixels" } ], "correctIndex": 3, "explanation": "DQN combined Q-learning with deep learning to learn Atari games directly from raw pixel data — no hand-coded rules or human demos — demonstrating the potential of a general AI system.", "uid": "1eaqkkm12eqqw8" }, { "id": "mcq-rw-lc-events", "shape": "mcq", "tags": [ "deepmind", "google", "ai-ethics" ], "prompt": { "modality": "text", "value": "Google's January 2014 acquisition of DeepMind — Europe's largest at the time (~£400 million) — came with an unusual condition. What was it?" }, "options": [ { "modality": "text", "value": "Barring DeepMind from taking any government contracts" }, { "modality": "text", "value": "Keeping all of DeepMind's research permanently closed-source" }, { "modality": "text", "value": "An independent ethics board to oversee AI applications" }, { "modality": "text", "value": "Relocating DeepMind's headquarters to California" } ], "correctIndex": 2, "explanation": "The deal, Europe's largest acquisition at the time at around £400 million, included a condition that an independent ethics board oversee how the AI was applied.", "uid": "1djpqps1am2dri" }, { "id": "cz-infinity-machine-def-deepmind", "shape": "cloze", "tags": [ "deepmind", "hassabis" ], "template": "___ is the London AI lab Demis Hassabis co-founded in 2010 and Google acquired in 2014. ___ stated mission is to solve intelligence and then use it to solve everything else.", "answer": "DeepMind", "distractors": [ "OpenAI", "Anthropic", "Cohere" ], "derivedFrom": "def-deepmind", "uid": "1bilf6x16525qf" }, { "id": "cz-infinity-machine-def-alphago", "shape": "cloze", "tags": [ "alphago", "milestones" ], "template": "___ is the DeepMind program that defeated world champion Lee Sedol at the board game Go in 2016. ___ combined deep neural networks with Monte Carlo tree search.", "answer": "AlphaGo", "distractors": [ "AlphaFold", "Deep Blue", "Watson" ], "derivedFrom": "def-alphago", "explanation": "Compare: AlphaFold — AlphaFold is the DeepMind system that predicts a protein's three-dimensional structure from its amino-acid sequence. AlphaFold cracked a fifty-year grand challenge in biology and helped win Hassabis a share of the 2024 Nobel Prize in Chemistry.", "uid": "1h8tfva18gcx5g" }, { "id": "cz-infinity-machine-def-alphafold", "shape": "cloze", "tags": [ "alphafold", "milestones" ], "template": "___ is the DeepMind system that predicts a protein's three-dimensional structure from its amino-acid sequence. ___ cracked a fifty-year grand challenge in biology and helped win Hassabis a share of the 2024 Nobel Prize in Chemistry.", "answer": "AlphaFold", "distractors": [ "AlphaGo", "AlphaZero", "Watson" ], "derivedFrom": "def-alphafold", "explanation": "Compare: AlphaGo — AlphaGo is the DeepMind program that defeated world champion Lee Sedol at the board game Go in 2016. AlphaGo combined deep neural networks with Monte Carlo tree search.", "uid": "phiy2o1vlvtnq" }, { "id": "cz-infinity-machine-def-reinforcement-learning", "shape": "cloze", "tags": [ "atari", "dqn" ], "template": "___ is a training method in which an agent learns by trial and error, receiving rewards for good actions and penalties for bad ones. DeepMind used ___ to master Atari games directly from raw pixels.", "answer": "Reinforcement learning", "distractors": [ "Supervised learning", "Unsupervised learning", "Transfer learning" ], "derivedFrom": "def-reinforcement-learning", "uid": "nb14bk21ecau" }, { "id": "cz-infinity-machine-def-gemini", "shape": "cloze", "tags": [ "gemini", "google" ], "template": "___ is Google DeepMind's flagship family of multimodal AI models, launched in 2023. ___ was the lab's answer to OpenAI's GPT-4 and marked its return to the research frontier.", "answer": "Gemini", "distractors": [ "Claude", "LLaMA", "PaLM" ], "derivedFrom": "def-gemini", "uid": "1ema2f3bc1yax" }, { "id": "cz-infinity-machine-def-elixir-studios", "shape": "cloze", "tags": [ "elixir", "biography" ], "template": "___ was the games company Hassabis founded after leaving Bullfrog. ___ built ambitious AI-driven games but struggled commercially and closed in 2005.", "answer": "Elixir Studios", "distractors": [ "Lionhead Studios", "Rockstar Games", "Rare" ], "derivedFrom": "def-elixir-studios", "uid": "1dt21c1r9g7nu" }, { "id": "cz-infinity-machine-def-casp", "shape": "cloze", "tags": [ "alphafold", "milestones" ], "template": "___ is the biennial Critical Assessment of Structure Prediction contest, where protein-folding methods compete on unsolved structures. AlphaFold's runaway victory at ___ in 2020 stunned the field of biology.", "answer": "CASP", "distractors": [ "ImageNet Challenge", "Netflix Prize", "DARPA Grand Challenge" ], "derivedFrom": "def-casp", "uid": "10o290fpglvo1" }, { "id": "cz-infinity-machine-fact-thesis-statement", "shape": "cloze", "tags": [ "intro", "agi", "race" ], "template": "Mallaby's argument: the most consequential technology of the century is being built by a small group of people whose biographies and rivalries matter — and DeepMind, born of a London poker game and a ___ pitch, has had outsized influence on what gets built and how.", "answer": "Singularity Summit", "distractors": [ "NeurIPS", "Davos", "TED" ], "derivedFrom": "fact-thesis-statement", "uid": "1jpe85o1nn2m8y" }, { "id": "tf-f-infinity-machine-pair-author", "shape": "trueFalse", "tags": [ "intro", "vocabulary" ], "statement": "Author of The Infinity Machine is Demis Hassabis.", "isTrue": false, "why": "Author of The Infinity Machine actually is Sebastian Mallaby.", "derivedFrom": "pair-author", "uid": "xhd924l1qk0y" }, { "id": "tf-t-infinity-machine-pair-subject", "shape": "trueFalse", "tags": [ "intro", "hassabis", "vocabulary" ], "statement": "\"Central subject of the book\" is Demis Hassabis.", "isTrue": true, "why": "Central subject of the book is Demis Hassabis.", "derivedFrom": "pair-subject", "uid": "yryzv9l5f1rn" }, { "id": "tf-f-infinity-machine-pair-mother", "shape": "trueFalse", "tags": [ "hassabis", "biography", "vocabulary" ], "statement": "\"Hassabis's mother's heritage\" is Greek Cypriot.", "isTrue": false, "why": "Hassabis's mother's heritage actually is Chinese Singaporean.", "derivedFrom": "pair-mother", "uid": "mnfpqi18gdejk" }, { "id": "tf-t-infinity-machine-pair-father", "shape": "trueFalse", "tags": [ "hassabis", "biography", "vocabulary" ], "statement": "\"Hassabis's father's heritage\" is Greek Cypriot.", "isTrue": true, "why": "Hassabis's father's heritage is Greek Cypriot.", "derivedFrom": "pair-father", "uid": "yo0ut4ry0eeq" }, { "id": "tf-f-infinity-machine-pair-bullfrog", "shape": "trueFalse", "tags": [ "hassabis", "biography", "vocabulary" ], "statement": "\"Peter Molyneux's game studio\" refers to Greek Cypriot.", "isTrue": false, "why": "Peter Molyneux's game studio actually refers to Bullfrog.", "derivedFrom": "pair-bullfrog", "uid": "de5bg9o1pn0b" }, { "id": "tf-t-infinity-machine-pair-idsia", "shape": "trueFalse", "tags": [ "legg", "places", "vocabulary" ], "statement": "Swiss AI lab where Legg studied with Hutter is \"IDSIA\".", "isTrue": true, "why": "Swiss AI lab where Legg studied with Hutter is IDSIA.", "derivedFrom": "pair-idsia", "uid": "p99tl1yz2s9f" }, { "id": "tf-f-infinity-machine-pair-moose", "shape": "trueFalse", "tags": [ "suleyman", "vocabulary" ], "statement": "\"Suleyman's nickname inside DeepMind\" refers to Geoffrey Irving.", "isTrue": false, "why": "Suleyman's nickname inside DeepMind actually refers to Moose.", "derivedFrom": "pair-moose", "uid": "qvq3301iy9gfy" }, { "id": "tf-t-infinity-machine-pair-aja", "shape": "trueFalse", "tags": [ "alphago", "people", "vocabulary" ], "statement": "Engineer who built AlphaGo's first prototype is Aja Huang.", "isTrue": true, "why": "Engineer who built AlphaGo's first prototype is Aja Huang.", "derivedFrom": "pair-aja", "uid": "1n251ynlza5dt" }, { "id": "tf-f-infinity-machine-pair-altman", "shape": "trueFalse", "tags": [ "openai", "altman", "people", "vocabulary" ], "statement": "OpenAI cofounder who became CEO is Ilya Sutskever.", "isTrue": false, "why": "OpenAI cofounder who became CEO actually is Sam Altman.", "derivedFrom": "pair-altman", "uid": "x7e42c1b6b376" }, { "id": "tf-t-infinity-machine-pair-sutskever", "shape": "trueFalse", "tags": [ "openai", "people", "vocabulary" ], "statement": "OpenAI's first chief scientist is Ilya Sutskever.", "isTrue": true, "why": "OpenAI's first chief scientist is Ilya Sutskever.", "derivedFrom": "pair-sutskever", "uid": "1cw46il1jnkd9j" }, { "id": "tf-t-infinity-machine-pair-hassabis-q", "shape": "trueFalse", "tags": [ "deepmind", "hassabis", "vocabulary" ], "statement": "\"DeepMind cofounder who won the 2024 Nobel Prize in Chemistry\" is Demis Hassabis.", "isTrue": true, "why": "DeepMind cofounder who won the 2024 Nobel Prize in Chemistry is Demis Hassabis.", "derivedFrom": "pair-hassabis-q", "uid": "19bwkmwplx8u" }, { "id": "tf-f-infinity-machine-pair-legg-q", "shape": "trueFalse", "tags": [ "deepmind", "legg", "vocabulary" ], "statement": "\"Cofounder who coined 'AGI'\" is Geoffrey Irving.", "isTrue": false, "why": "Cofounder who coined \"AGI\" actually is Shane Legg.", "derivedFrom": "pair-legg-q", "uid": "18uzipkbwi16u" }, { "id": "tf-t-infinity-machine-pair-suleyman-q", "shape": "trueFalse", "tags": [ "deepmind", "suleyman", "vocabulary" ], "statement": "\"Cofounder who became Microsoft AI CEO\" is Mustafa Suleyman.", "isTrue": true, "why": "Cofounder who became Microsoft AI CEO is Mustafa Suleyman.", "derivedFrom": "pair-suleyman-q", "uid": "mzptlj1156289" }, { "id": "tf-f-infinity-machine-pair-silver-q", "shape": "trueFalse", "tags": [ "deepmind", "alphago", "vocabulary" ], "statement": "\"Cambridge friend who led AlphaGo\" is Mustafa Suleyman.", "isTrue": false, "why": "Cambridge friend who led AlphaGo actually is David Silver.", "derivedFrom": "pair-silver-q", "uid": "eyuuty18in2jw" }, { "id": "tf-f-infinity-machine-pair-jumper-q", "shape": "trueFalse", "tags": [ "deepmind", "alphafold", "vocabulary" ], "statement": "\"Scientist who led AlphaFold 2\" is Vlad Mnih.", "isTrue": false, "why": "Scientist who led AlphaFold 2 actually is John Jumper.", "derivedFrom": "pair-jumper-q", "uid": "qkvi2n1xzj4xt" }, { "id": "tf-t-infinity-machine-pair-thiel-q", "shape": "trueFalse", "tags": [ "policy", "thiel", "vocabulary" ], "statement": "\"First investor in DeepMind\" is Peter Thiel.", "isTrue": true, "why": "First investor in DeepMind is Peter Thiel.", "derivedFrom": "pair-thiel-q", "uid": "vcrt361lw8qys" }, { "id": "tf-t-infinity-machine-q-im-deepmind-firstname", "shape": "trueFalse", "tags": [ "deepmind", "vocabulary" ], "statement": "\"DeepMind's first working name\" refers to \"Solaria, taken from the Asimov novel\".", "isTrue": true, "why": "DeepMind's first working name refers to Solaria, taken from the Asimov novel.", "derivedFrom": "q-im-deepmind-firstname", "uid": "1cst6szuri5u5" }, { "id": "tf-f-infinity-machine-q-im-openai-cap", "shape": "trueFalse", "tags": [ "openai", "altman", "governance" ], "statement": "\"OpenAI's 2019 investor return cap\" refers to \"Undisclosed Gulf fundraising and pushing out critical board members\".", "isTrue": false, "why": "OpenAI's 2019 investor return cap actually refers to Profit capped at 100 times for early investors.", "derivedFrom": "q-im-openai-cap", "uid": "1edb194u12cgu" }, { "id": "tf-f-infinity-machine-q-im-google-deepmind-merger", "shape": "trueFalse", "tags": [ "google", "deepmind", "race" ], "statement": "\"April 2023 merger that formed Google DeepMind\" refers to \"At a dinner at Larry Page's home in October 2012\".", "isTrue": false, "why": "April 2023 merger that formed Google DeepMind actually refers to Google Brain merged into DeepMind under Hassabis.", "derivedFrom": "q-im-google-deepmind-merger", "uid": "h73y8b1wi5ujd" }, { "id": "tf-f-infinity-machine-q-im-buchanan-czar", "shape": "trueFalse", "tags": [ "buchanan", "governance", "milestones" ], "statement": "\"Biden's June 2023 AI czar\" refers to \"At the White House, with no major US frontier lab declining\".", "isTrue": false, "why": "Biden's June 2023 AI czar actually refers to Ben Buchanan, a Georgetown AI and cybersecurity professor.", "derivedFrom": "q-im-buchanan-czar", "uid": "1n0eni5t476er" }, { "id": "tf-f-infinity-machine-q-im-gemini15-release", "shape": "trueFalse", "tags": [ "gemini", "milestones" ], "statement": "\"Gemini 1.5 Pro's headline capability at launch\" refers to \"At the White House, with no major US frontier lab declining\".", "isTrue": false, "why": "Gemini 1.5 Pro's headline capability at launch actually refers to A million-token context window with a mixture-of-experts design.", "derivedFrom": "q-im-gemini15-release", "uid": "p5phv6bvwv0s" }, { "id": "tf-f-infinity-machine-q-im-altman-firing-triggers", "shape": "trueFalse", "tags": [ "openai", "altman" ], "statement": "\"Two triggers Mallaby gives for the Altman firing\" refers to \"Profit capped at 100 times for early investors\".", "isTrue": false, "why": "Two triggers Mallaby gives for the Altman firing actually refers to Undisclosed Gulf fundraising and pushing out critical board members.", "derivedFrom": "q-im-altman-firing-triggers", "uid": "1up7ec5r2vfer" }, { "id": "tf-f-infinity-machine-q-im-hassabis-nobel", "shape": "trueFalse", "tags": [ "intro", "race", "hassabis" ], "statement": "\"Hassabis's dual status as the book closes in 2025-26\" refers to Sebastian Mallaby.", "isTrue": false, "why": "Hassabis's dual status as the book closes in 2025-26 actually refers to A Nobel laureate and chief executive of a leading AI lab.", "derivedFrom": "q-im-hassabis-nobel", "uid": "ypizxw7k1pce" }, { "id": "tf-d-infinity-machine-def-agi", "shape": "trueFalse", "tags": [ "intro", "agi", "vocabulary" ], "statement": "AGI: Artificial general intelligence — a system that can match or surpass human intelligence across the full range of cognitive tasks, not just one narrow domain.", "isTrue": true, "why": "That definition is correct for \"AGI\".", "derivedFrom": "def-agi", "uid": "1cakzqa1c9jk4k" }, { "id": "tf-d-infinity-machine-def-dqn", "shape": "trueFalse", "tags": [ "dqn", "atari", "vocabulary" ], "statement": "DQN (Deep Q-Network): A reinforcement-learning architecture that pairs Q-learning — predicting the long-term value of each possible action — with a deep convolutional network that reads raw pixels.", "isTrue": true, "why": "That definition is correct for \"DQN (Deep Q-Network)\".", "derivedFrom": "def-dqn", "uid": "125jl5n9z34gp" }, { "id": "tf-df-infinity-machine-def-q-learning", "shape": "trueFalse", "tags": [ "dqn", "vocabulary" ], "statement": "DQN (Deep Q-Network): A reinforcement-learning algorithm in which an agent learns the expected long-term reward of taking each action from each state.", "isTrue": false, "why": "That's the definition of \"Q-learning\", not \"DQN (Deep Q-Network)\".", "derivedFrom": "def-q-learning", "uid": "19sq6paahyfxs" }, { "id": "tf-d-infinity-machine-def-protein-folding", "shape": "trueFalse", "tags": [ "alphafold", "vocabulary" ], "statement": "Protein folding: The process by which a chain of amino acids twists itself into a specific three-dimensional shape that determines a protein's function.", "isTrue": true, "why": "That definition is correct for \"Protein folding\".", "derivedFrom": "def-protein-folding", "uid": "7x3tiy1xjzeoc" }, { "id": "tf-d-infinity-machine-def-gdt", "shape": "trueFalse", "tags": [ "alphafold", "vocabulary" ], "statement": "GDT (Global Distance Test): CASP's headline accuracy score.", "isTrue": true, "why": "That definition is correct for \"GDT (Global Distance Test)\".", "derivedFrom": "def-gdt", "uid": "m6gljjbvldnt" }, { "id": "tf-d-infinity-machine-def-alignment", "shape": "trueFalse", "tags": [ "safety", "vocabulary" ], "statement": "Alignment: The technical and conceptual problem of ensuring that an AI system reliably pursues human-intended goals, even as the system becomes more capable.", "isTrue": true, "why": "That definition is correct for \"Alignment\".", "derivedFrom": "def-alignment", "uid": "1tu5heu1xaobmg" }, { "id": "tf-d-infinity-machine-def-mmlu", "shape": "trueFalse", "tags": [ "gemini", "vocabulary" ], "statement": "MMLU (Massive Multitask Language Understanding): A benchmark of 57 subjects spanning maths, humanities, ethics and law.", "isTrue": true, "why": "That definition is correct for \"MMLU (Massive Multitask Language Understanding)\".", "derivedFrom": "def-mmlu", "uid": "1p403yabz5r3k" }, { "id": "tf-df-infinity-machine-def-deepmind", "shape": "trueFalse", "tags": [ "deepmind", "hassabis" ], "statement": "AGI: The London AI lab Demis Hassabis co-founded in 2010 and Google acquired in 2014.", "isTrue": false, "why": "That's the definition of \"DeepMind\", not \"AGI\".", "derivedFrom": "def-deepmind", "uid": "1y3yrhncr6vil" }, { "id": "tf-df-infinity-machine-def-alphago", "shape": "trueFalse", "tags": [ "alphago", "milestones" ], "statement": "CASP: The DeepMind program that defeated world champion Lee Sedol at the board game Go in 2016.", "isTrue": false, "why": "That's the definition of \"AlphaGo\", not \"CASP\".", "derivedFrom": "def-alphago", "uid": "5lyuv51mjob9z" }, { "id": "tf-d-infinity-machine-def-alphafold", "shape": "trueFalse", "tags": [ "alphafold", "milestones" ], "statement": "AlphaFold: The DeepMind system that predicts a protein's three-dimensional structure from its amino-acid sequence.", "isTrue": true, "why": "That definition is correct for \"AlphaFold\".", "derivedFrom": "def-alphafold", "uid": "17w7l1717jqejp" }, { "id": "tf-df-infinity-machine-def-reinforcement-learning", "shape": "trueFalse", "tags": [ "atari", "dqn" ], "statement": "DQN (Deep Q-Network): A training method in which an agent learns by trial and error, receiving rewards for good actions and penalties for bad ones.", "isTrue": false, "why": "That's the definition of \"Reinforcement learning\", not \"DQN (Deep Q-Network)\".", "derivedFrom": "def-reinforcement-learning", "uid": "duh5j31y1kdzp" }, { "id": "tf-d-infinity-machine-def-gemini", "shape": "trueFalse", "tags": [ "gemini", "google" ], "statement": "Gemini: Google DeepMind's flagship family of multimodal AI models, launched in 2023.", "isTrue": true, "why": "That definition is correct for \"Gemini\".", "derivedFrom": "def-gemini", "uid": "qot4jm13g1614" }, { "id": "tf-d-infinity-machine-def-elixir-studios", "shape": "trueFalse", "tags": [ "elixir", "biography" ], "statement": "Elixir Studios: The games company Hassabis founded after leaving Bullfrog.", "isTrue": true, "why": "That definition is correct for \"Elixir Studios\".", "derivedFrom": "def-elixir-studios", "uid": "12tiysy1dsuhoo" }, { "id": "tf-d-infinity-machine-def-casp", "shape": "trueFalse", "tags": [ "alphafold", "milestones" ], "statement": "CASP: The biennial Critical Assessment of Structure Prediction contest, where protein-folding methods compete on unsolved structures.", "isTrue": true, "why": "That definition is correct for \"CASP\".", "derivedFrom": "def-casp", "uid": "1w6kdj7t3j3it" }, { "id": "concept-rw-infinity-machine-peter-thiel", "shape": "concept", "conceptKind": "person", "tags": [ "person", "funding", "investors" ], "name": "Peter Thiel", "clues": [ "An early investor who anchored DeepMind's very first funding round.", "He backed the company through Founders Fund after its 2010 Singularity Summit pitch.", "Walking to the car after that pitch, he reportedly muttered \"Demis's destiny.\"" ], "uid": "guwntk1xfpthq" }, { "id": "concept-rw-infinity-machine-sam-altman", "shape": "concept", "conceptKind": "person", "tags": [ "person", "openai", "rivalry" ], "name": "Sam Altman", "clues": [ "The chief executive of a rival lab at the center of the AI race.", "He led OpenAI, and Mallaby treats a crisis around him as the death of any institutional brake on the race.", "Fired on 17 November 2023, he was reinstated just five days later." ], "uid": "17rsvw7emzhgt" }, { "id": "concept-rw-infinity-machine-ilya-sutskever", "shape": "concept", "conceptKind": "person", "tags": [ "person", "openai", "rivalry" ], "name": "Ilya Sutskever", "clues": [ "A leading AI scientist who helped engineer a dramatic corporate revolt.", "As OpenAI's chief scientist, he was central to the board's move against its CEO.", "He reversed himself once Microsoft offered to absorb the company." ], "uid": "14a3uo1ixu9wi" }, { "id": "concept-rw-infinity-machine-geoffrey-irving", "shape": "concept", "conceptKind": "person", "tags": [ "person", "ai-safety" ], "name": "Geoffrey Irving", "clues": [ "A researcher focused on making AI systems safe rather than merely capable.", "He built DeepMind's alignment programme.", "He later joined the UK AI Safety Institute." ], "uid": "cbpbbi1s9foqg" }, { "id": "concept-rw-infinity-machine-ben-buchanan", "shape": "concept", "conceptKind": "person", "tags": [ "person", "ai-policy" ], "name": "Ben Buchanan", "clues": [ "A government official who shaped US policy on artificial intelligence.", "He served as the Biden White House's AI czar.", "He was the architect of the July 2023 voluntary commitments and the October 2023 executive order." ], "uid": "12s4j1bqaru11" }, { "id": "concept-rw-infinity-machine-the-sweetness", "shape": "concept", "conceptKind": "thing", "tags": [ "thing", "metaphor", "discovery" ], "name": "The sweetness", "clues": [ "Hassabis's private name for a particular internal signal, and Mallaby's organising metaphor for the whole book.", "It arrives at the instant a long-resistant problem finally breaks open — the moment before any announcement, alone with the solution.", "Not the prize or the paper but this feeling, Mallaby argues, is why anyone pours a decade into a single research problem." ], "uid": "1416xybzk0znx" }, { "id": "concept-rw-infinity-machine-the-infinity-machine", "shape": "concept", "conceptKind": "thing", "tags": [ "thing", "agi", "metaphor" ], "name": "The infinity machine", "clues": [ "Mallaby's deliberately grandiose name for the kind of system DeepMind has always aimed to build.", "His label for a mind that could, in principle, solve any problem a human can frame — the full cognitive range, not one narrow task.", "The book's central metaphor: a constructed intelligence the author suggests may have no upper bound." ], "uid": "rwj96cyiznni" }, { "id": "concept-rw-infinity-machine-nobel-prize-in-chemistry", "shape": "concept", "conceptKind": "thing", "tags": [ "thing", "award", "alphafold" ], "name": "Nobel Prize in Chemistry", "clues": [ "A top international science award, presented in one of its scientific categories.", "The 2024 honour that Mallaby's book opens on, with Hassabis stepping up to collect it.", "It was awarded for AlphaFold's prediction of protein structures from amino-acid sequences." ], "uid": "1i851kqnz1u8" }, { "id": "concept-rw-infinity-machine-zx-spectrum", "shape": "concept", "conceptKind": "thing", "tags": [ "thing", "computer", "childhood" ], "name": "ZX Spectrum", "clues": [ "The home computer an eight-year-old Hassabis bought with his chess winnings.", "On it he taught himself BASIC and machine code straight from the manuals.", "Within a year he used it to write a primitive neural network — what the book calls a straight line to AlphaZero." ], "uid": "94gdfk1dclvey" }, { "id": "concept-rw-infinity-machine-theme-park", "shape": "concept", "conceptKind": "thing", "tags": [ "thing", "game", "bullfrog" ], "name": "Theme Park", "clues": [ "The management sim the teenage Hassabis co-designed at Bullfrog.", "He was co-designer and lead programmer on it, working under Peter Molyneux.", "This management simulation went on to sell millions of copies." ], "uid": "acu3k71sk9pgd" }, { "id": "concept-rw-infinity-machine-g-del-escher-bach", "shape": "concept", "conceptKind": "thing", "tags": [ "thing", "book", "hofstadter" ], "name": "Gödel, Escher, Bach", "clues": [ "Mallaby singles it out as Hassabis's most-loved book of his teens.", "Douglas Hofstadter's braided account of self-reference, formal systems and strange loops of mind.", "It gave Hassabis his first articulate sense that intelligence might be built from layered patterns rather than something mystical." ], "uid": "1b331161pgr34w" }, { "id": "concept-rw-infinity-machine-bullfrog-productions", "shape": "concept", "conceptKind": "thing", "tags": [ "thing", "studio", "games" ], "name": "Bullfrog Productions", "clues": [ "The Guildford game studio a seventeen-year-old Hassabis walked into looking for work before Cambridge.", "Peter Molyneux's outfit, where Hassabis became a co-designer and lead programmer.", "The company behind Theme Park — the place Mallaby argues was Hassabis's real early career." ], "uid": "1e4tdp9hu2zuv" }, { "id": "concept-rw-infinity-machine-go", "shape": "concept", "conceptKind": "thing", "tags": [ "thing", "board-game", "cambridge" ], "name": "Go", "clues": [ "A board game two computer-science freshers bonded over in their college's Junior Common Room in 1994.", "The shared passion of Hassabis and David Silver that sparked the longest partnership of Hassabis's career.", "This board game later became the subject of the celebrated DeepMind project David Silver led." ], "uid": "gvbees15hp9ra" }, { "id": "concept-rw-infinity-machine-hippocampus", "shape": "concept", "conceptKind": "thing", "tags": [ "thing", "neuroscience", "brain" ], "name": "hippocampus", "clues": [ "A brain region Hassabis studied during his cognitive-neuroscience doctorate at UCL under Eleanor Maguire.", "He found that patients with damage to it could not imagine future scenes.", "His 2007 PNAS finding: the same structure that stores memories also constructs hypothetical futures." ], "uid": "xnac3g72h0h2" }, { "id": "concept-rw-infinity-machine-generative-model", "shape": "concept", "conceptKind": "thing", "tags": [ "thing", "neuroscience", "ai" ], "name": "generative model", "clues": [ "Hassabis's core conclusion about what the brain fundamentally is.", "Not a lookup table, he argued, but something that actively constructs — an idea that quietly structured DeepMind's research agenda.", "In machine-learning terms, the two-word name for a system that produces new data rather than merely retrieving stored answers." ], "uid": "p9wj14n9dgu2" }, { "id": "concept-rw-infinity-machine-marcus-hutter", "shape": "concept", "conceptKind": "person", "tags": [ "person", "researcher", "agi" ], "name": "Marcus Hutter", "clues": [ "An AI theorist based at IDSIA, the Swiss institute in Lugano.", "He and Shane Legg developed theories of universal intelligence together.", "Around 2002 he and Legg were brainstorming a book title when Legg proposed the phrase 'Artificial General Intelligence.'" ], "uid": "pfavt8p24hfu" }, { "id": "concept-rw-infinity-machine-j-rgen-schmidhuber", "shape": "concept", "conceptKind": "person", "tags": [ "person", "researcher", "neural-networks" ], "name": "Jürgen Schmidhuber", "clues": [ "A researcher at the Swiss AI institute IDSIA in Lugano.", "Shane Legg worked with him there on recurrent neural networks.", "A pioneer of recurrent neural networks, he was among Legg's collaborators before Legg helped found DeepMind." ], "uid": "qkc4szvgb9zp" }, { "id": "concept-rw-infinity-machine-george-hassabis", "shape": "concept", "conceptKind": "person", "tags": [ "person", "family", "founding" ], "name": "George Hassabis", "clues": [ "The family link through whom Demis first met Mustafa Suleyman.", "Demis's brother, who had been at school with DeepMind's future operations man.", "It was through this sibling that Demis met the operator who became the lab's third founding leg." ], "uid": "1slutbv9qb10t" }, { "id": "concept-rw-infinity-machine-luke-nosek", "shape": "concept", "conceptKind": "person", "tags": [ "person", "investor", "founders-fund" ], "name": "Luke Nosek", "clues": [ "A partner at Peter Thiel's venture firm.", "He helped walk the three DeepMind founders back to their car after the Thiel breakfast.", "The Founders Fund partner to whom Thiel muttered 'Demis's destiny.'" ], "uid": "6rcfr5zh48m3" }, { "id": "concept-rw-infinity-machine-isaac-asimov", "shape": "concept", "conceptKind": "person", "tags": [ "person", "author", "naming" ], "name": "Isaac Asimov", "clues": [ "A science-fiction author whose novel supplied DeepMind's first working name.", "The company was briefly called Solaria, borrowed from one of his books.", "This SF writer's work gave the startup its placeholder name before the founders settled on 'Deep' plus 'Mind.'" ], "uid": "la7ukoj8ybmu" }, { "id": "concept-rw-infinity-machine-larry-page", "shape": "concept", "conceptKind": "person", "tags": [ "person", "google", "acquisition" ], "name": "Larry Page", "clues": [ "A Google co-founder who had been watching DeepMind since 2012.", "He hosted the October 2012 Palo Alto dinner where the future of AI was argued.", "He countered Musk by insisting the species that built a superintelligence would be its natural heir." ], "uid": "1bj8sy610nbg60" }, { "id": "concept-rw-infinity-machine-elon-musk", "shape": "concept", "conceptKind": "person", "tags": [ "person", "ai-safety", "openai" ], "name": "Elon Musk", "clues": [ "A guest at Larry Page's 2012 Palo Alto dinner who clashed with the host over AI.", "He warned that a sufficiently advanced AI could become 'a digital god,' with no plan for the aftermath.", "Mallaby frames his split with Page that night as the seed of what became OpenAI." ], "uid": "v99anthhphmv" }, { "id": "concept-rw-infinity-machine-fan-hui", "shape": "concept", "conceptKind": "person", "tags": [ "person", "alphago", "go" ], "name": "Fan Hui", "clues": [ "A reigning European champion of an ancient Asian board game, chosen for AI's first serious professional test.", "In a closed-door five-game match in October 2015, he was beaten without winning a single game.", "Moved by the loss, he joined DeepMind as an analyst; the 5-0 result stayed under embargo until a January 2016 Nature cover." ], "uid": "1jq231c182cj2y" }, { "id": "concept-rw-infinity-machine-lee-sedol", "shape": "concept", "conceptKind": "person", "tags": [ "person", "alphago", "go" ], "name": "Lee Sedol", "clues": [ "Widely regarded as the greatest player of his generation at an ancient board game, he was AI's next challenge after a European champion fell.", "In a March 2016 Seoul match watched by tens of millions across Asia, he lost four games to one.", "His lone victory, game four, exploited a move the machine had failed to evaluate; earlier he had left the board shaken by its move 37." ], "uid": "yl56g2hhv8n0" }, { "id": "concept-rw-infinity-machine-aja-huang", "shape": "concept", "conceptKind": "person", "tags": [ "person", "alphago", "deepmind" ], "name": "Aja Huang", "clues": [ "A DeepMind engineer who built the very first prototype of the lab's Go-playing system.", "His work on the tree-search component proved critical to making that system competitive at professional level." ], "uid": "125u26i18fblu8" }, { "id": "concept-rw-infinity-machine-sergey-brin", "shape": "concept", "conceptKind": "person", "tags": [ "person", "google", "alphago" ], "name": "Sergey Brin", "clues": [ "A Google co-founder who, soon after the DeepMind acquisition, asked the lab's leader what hard problem it would tackle next.", "On a May 2014 hallway walk he heard the answer 'Go' and was sceptical of a game with about 10^170 positions.", "Despite his doubts he provided the resources, and the project he funded launched that summer." ], "uid": "1fd0f937iyjhx" }, { "id": "concept-rw-infinity-machine-greg-brockman", "shape": "concept", "conceptKind": "person", "tags": [ "person", "openai" ], "name": "Greg Brockman", "clues": [ "On 20 September 2017 this OpenAI co-founder sent a warning memo to the lab's backers.", "The memo, co-signed with the chief scientist, warned that one person could gain 'absolute control' of AI.", "Its unnamed target of concern was the head of a rival London lab." ], "uid": "tibm6u1tpr7z8" }, { "id": "concept-rw-infinity-machine-dario-amodei", "shape": "concept", "conceptKind": "person", "tags": [ "person", "anthropic" ], "name": "Dario Amodei", "clues": [ "OpenAI's VP of Research who came to believe safety demanded a clean break.", "In 2021 he and his sister led roughly a dozen researchers out of OpenAI.", "The lab they founded organised itself around Constitutional AI and interpretability rather than raw capability." ], "uid": "19j494o1yv5afy" }, { "id": "concept-rw-infinity-machine-satya-nadella", "shape": "concept", "conceptKind": "person", "tags": [ "person", "microsoft" ], "name": "Satya Nadella", "clues": [ "The CEO of a cloud giant that had become the silent guarantor of OpenAI's continuity.", "Within 36 hours of a rival CEO's firing, he announced his company would build that ousted leader a new AI lab.", "He offered a job to any of the 700-plus employees threatening to walk, making the board's position untenable." ], "uid": "1nzylej1ursa4d" }, { "id": "concept-rw-infinity-machine-anfinsen-s-conjecture", "shape": "concept", "conceptKind": "thing", "tags": [ "thing", "alphafold", "protein-folding" ], "name": "Anfinsen's conjecture", "clues": [ "An idea from a 1972 Nobel lecture about what determines a protein's final shape.", "It held that a chain's amino-acid sequence alone dictates the folded form, needing no extra cellular machinery.", "Biologists accepted it for half a century but couldn't compute it, until AlphaFold vindicated it roughly fifty years later." ], "uid": "1o70r8vvicqkt" }, { "id": "concept-rw-infinity-machine-casp", "shape": "concept", "conceptKind": "thing", "tags": [ "thing", "alphafold", "protein-folding" ], "name": "CASP", "clues": [ "A biennial contest, running since 1994, that pits protein-folding methods against one another.", "Run by Professor John Moult, it hands labs amino-acid sequences whose true shapes only the experimentalists know.", "It gave AlphaFold an external scoreboard, from a 2018 Cancun edition to a landmark 2020 edition where the score hit 92 GDT." ], "uid": "1ftow9c16is4lu" }, { "id": "concept-rw-infinity-machine-gdt-global-distance-test", "shape": "concept", "conceptKind": "thing", "tags": [ "thing", "alphafold", "metric" ], "name": "GDT (Global Distance Test)", "clues": [ "The headline accuracy score used to judge protein-structure predictions.", "It measures how closely predicted atom positions match the experimentally verified ones.", "During a risky architectural pivot it crashed from 60 to 20; by CASP14 it reached 92, over 50% above any prior best." ], "uid": "1p4drnz19ghx69" }, { "id": "concept-rw-infinity-machine-distogram", "shape": "concept", "conceptKind": "thing", "tags": [ "thing", "alphafold", "architecture" ], "name": "distogram", "clues": [ "AlphaFold 1's defining output, replacing a simple yes/no prediction about pairs of amino acids.", "Instead of asking whether two residues touch, it predicted the continuous distance between them.", "A team member likened the richer geometric picture to going from black-and-white to full-colour television." ], "uid": "4inqyn1h55dkx" }, { "id": "concept-rw-infinity-machine-yoshua-bengio", "shape": "concept", "conceptKind": "person", "tags": [ "person", "ai-safety" ], "name": "Yoshua Bengio", "clues": [ "A pioneering deep-learning researcher who lent his name to a stark 2023 warning about AI.", "He signed a one-sentence statement ranking AI extinction risk alongside pandemics and nuclear war.", "At the Bletchley Park summit he pressed DeepMind's safety advisers on how fast they were deploying." ], "uid": "a76j8s8whv82" }, { "id": "concept-rw-infinity-machine-yann-lecun", "shape": "concept", "conceptKind": "person", "tags": [ "person", "ai-safety" ], "name": "Yann LeCun", "clues": [ "A prominent AI pioneer who broke with peers over doomsday framing.", "When a one-sentence statement warned of AI extinction risk, he pointedly declined to sign.", "His refusal reflected a long-held scepticism about near-term catastrophic risk." ], "uid": "1jkb641ahh1aa" }, { "id": "concept-rw-infinity-machine-ke-jie", "shape": "concept", "conceptKind": "person", "tags": [ "person", "go", "china" ], "name": "Ke Jie", "clues": [ "A reigning world champion of an ancient board game, defeated by AlphaGo in 2017.", "Inside Beijing his loss was called a 'Sputnik moment', proof of a Western technological lead.", "The defeat helped spur China's State Council toward a formal AI strategy with 2030 targets." ], "uid": "1rxvx4r15f2x0d" }, { "id": "concept-rw-infinity-machine-gemini-ultra", "shape": "concept", "conceptKind": "thing", "tags": [ "thing", "gemini", "benchmark" ], "name": "Gemini Ultra", "clues": [ "The headline variant of Google DeepMind's first frontier model built jointly by its London and Mountain View teams.", "Launched 6 December 2023, it was previewed head-to-head against GPT-4.", "Its announced score of 90% on a 57-subject benchmark edged past GPT-4's published number." ], "uid": "xqnjky1obdzd8" }, { "id": "concept-rw-infinity-machine-mmlu", "shape": "concept", "conceptKind": "thing", "tags": [ "thing", "benchmark" ], "name": "MMLU", "clues": [ "A benchmark used to preview Google DeepMind's flagship model against GPT-4.", "It spans 57 subjects, from maths and humanities to ethics and law.", "The London-Mountain View model's announced 90% on it edged past GPT-4's published number." ], "uid": "26vxxv1hn2gu1" }, { "id": "concept-rw-infinity-machine-gemini-1-5-pro", "shape": "concept", "conceptKind": "thing", "tags": [ "thing", "gemini" ], "name": "Gemini 1.5 Pro", "clues": [ "A February 2024 model release built by a London-led 'strike team' of the kind that had produced AlphaFold.", "Its headline capability was a million-token context window.", "It was the first release of Google DeepMind's flagship line where London, not Mountain View, set the research agenda." ], "uid": "1b4715v1v5fv39" }, { "id": "concept-rw-infinity-machine-sparse-mixture-of-experts", "shape": "concept", "conceptKind": "thing", "tags": [ "thing", "architecture" ], "name": "sparse mixture-of-experts", "clues": [ "An efficiency-minded architecture paired with a million-token context window in a February 2024 model.", "Only a fraction of the model's parameters activate for any given input, keeping inference costs manageable." ], "uid": "f702ru1tgpi9s" }, { "id": "ot-book-basics-1", "shape": "mcq", "tags": [ "author", "attribution" ], "prompt": { "modality": "text", "value": "Who wrote The Infinity Machine?" }, "options": [ { "modality": "text", "value": "Marcus Hutter" }, { "modality": "text", "value": "Shane Legg" }, { "modality": "text", "value": "Sebastian Mallaby" }, { "modality": "text", "value": "Demis Hassabis" } ], "correctIndex": 2, "explanation": "Sebastian Mallaby is the author, previously known for writing about hedge funds and venture capital. Demis Hassabis is the book's central subject, not its author.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "5ftdyo17uwea2" }, { "id": "ot-book-basics-2", "shape": "mcq", "tags": [ "subject", "hassabis" ], "prompt": { "modality": "text", "value": "Who is the central subject of The Infinity Machine?" }, "options": [ { "modality": "text", "value": "Demis Hassabis" }, { "modality": "text", "value": "Shane Legg" }, { "modality": "text", "value": "Sebastian Mallaby" }, { "modality": "text", "value": "Sam Altman" } ], "correctIndex": 0, "explanation": "The book centres on Demis Hassabis — chess prodigy, game designer, neuroscience PhD, Nobel laureate. Mallaby is the author, not the subject.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1b1wprh1ovohwv" }, { "id": "ot-book-basics-3", "shape": "mcq", "tags": [ "author", "background" ], "prompt": { "modality": "text", "value": "Before The Infinity Machine, Sebastian Mallaby was best known for writing about which subjects?" }, "options": [ { "modality": "text", "value": "Neuroscience and cognitive psychology" }, { "modality": "text", "value": "Chess and competitive gaming" }, { "modality": "text", "value": "Hedge funds and venture capital" }, { "modality": "text", "value": "Nuclear weapons policy" } ], "correctIndex": 2, "explanation": "Mallaby came to this book from a career writing about hedge funds and venture capital. Neuroscience and chess belong to Hassabis's biography, not Mallaby's.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "i2j7k11xvvcff" }, { "id": "ot-book-basics-4", "shape": "mcq", "tags": [ "attribution", "which-not" ], "prompt": { "modality": "text", "value": "Which statement about The Infinity Machine is NOT accurate?" }, "options": [ { "modality": "text", "value": "Penguin Press published it" }, { "modality": "text", "value": "It was published in 2026" }, { "modality": "text", "value": "Sebastian Mallaby is the book's author" }, { "modality": "text", "value": "Demis Hassabis is the book's author" } ], "correctIndex": 3, "explanation": "Hassabis is the central subject, not the author — Mallaby wrote the book. The other three statements are all correct: Penguin Press, 2026.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1vulyku56yzt0" }, { "id": "ot-sweetness-1", "shape": "mcq", "tags": [ "sweetness", "discovery" ], "prompt": { "modality": "text", "value": "In the book, what does Hassabis mean by 'the sweetness'?" }, "options": [ { "modality": "text", "value": "The instant a long-resistant problem finally breaks open" }, { "modality": "text", "value": "The moment a research grant is finally approved" }, { "modality": "text", "value": "The satisfaction of publishing ahead of a rival lab" }, { "modality": "text", "value": "The public ceremony where a discovery is honoured" } ], "correctIndex": 0, "explanation": "The sweetness is the private rush at the instant a problem cracks — explicitly not the public celebration or the announcement that follows.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1fp2isdqdp02v" }, { "id": "ot-sweetness-2", "shape": "mcq", "tags": [ "nobel", "opening-scene" ], "prompt": { "modality": "text", "value": "The scene Mallaby opens the book with is Hassabis collecting which award?" }, "options": [ { "modality": "text", "value": "The Nobel Prize in Chemistry" }, { "modality": "text", "value": "The Turing Award" }, { "modality": "text", "value": "The Nobel Prize in Physics" }, { "modality": "text", "value": "The Nobel Prize in Physiology or Medicine" } ], "correctIndex": 0, "explanation": "The book opens on Hassabis collecting the 2024 Nobel Prize in Chemistry, awarded for AlphaFold's protein-structure work.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1adbl4e16gnw5k" }, { "id": "ot-sweetness-3", "shape": "mcq", "tags": [ "alphafold", "nobel" ], "prompt": { "modality": "text", "value": "The Nobel that frames the book's opening was awarded for which achievement?" }, "options": [ { "modality": "text", "value": "Training an agent to play Atari games from raw pixels" }, { "modality": "text", "value": "Building the Gemini family of models" }, { "modality": "text", "value": "Defeating a world champion at the board game Go" }, { "modality": "text", "value": "Predicting protein structures from amino-acid sequences" } ], "correctIndex": 3, "explanation": "The 2024 Chemistry Nobel recognised AlphaFold's prediction of protein structures from amino-acid sequences. AlphaGo, the Atari DQN and Gemini are milestones, but not the Nobel work.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1np8q7v1b6p5gx" }, { "id": "ot-sweetness-4", "shape": "mcq", "tags": [ "sweetness", "metaphor" ], "prompt": { "modality": "text", "value": "Why does Mallaby make 'the sweetness' his organising metaphor?" }, "options": [ { "modality": "text", "value": "It shows that prizes are the true motive for research" }, { "modality": "text", "value": "It proves scientific discovery is mostly luck" }, { "modality": "text", "value": "It explains why researchers pour a decade into one problem" }, { "modality": "text", "value": "It argues that discovery is a purely romantic notion" } ], "correctIndex": 2, "explanation": "The metaphor explains motivation — the internal signal, not the prize or the paper, is what sustains a decade on one problem. Mallaby uses the Nobel precisely to show the feeling is not merely romantic.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1aeq7hr1lwhf7x" }, { "id": "ot-sweetness-5", "shape": "mcq", "tags": [ "sweetness", "which-not" ], "prompt": { "modality": "text", "value": "Which moment would Hassabis NOT count as 'the sweetness'?" }, "options": [ { "modality": "text", "value": "The private rush the instant an answer appears" }, { "modality": "text", "value": "Standing on stage as the prize is announced" }, { "modality": "text", "value": "Seeing a decade-old problem suddenly crack" }, { "modality": "text", "value": "Being alone with a solution before anyone knows" } ], "correctIndex": 1, "explanation": "The sweetness is the moment before the announcement, alone with the solution — a public stage is exactly what it is not.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "14ypq3110k2qdz" }, { "id": "ot-agi-1", "shape": "mcq", "tags": [ "agi", "title" ], "prompt": { "modality": "text", "value": "'The infinity machine' is Mallaby's label for what?" }, "options": [ { "modality": "text", "value": "Reinforcement learning" }, { "modality": "text", "value": "Artificial general intelligence" }, { "modality": "text", "value": "DeepMind's Gemini model family" }, { "modality": "text", "value": "The transformer architecture" } ], "correctIndex": 1, "explanation": "The title is Mallaby's metaphor for artificial general intelligence. Transformers, Gemini and reinforcement learning are components or milestones on the path, not the machine itself.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1bjbq2j2qvwnd" }, { "id": "ot-agi-2", "shape": "mcq", "tags": [ "agi", "definition" ], "prompt": { "modality": "text", "value": "How does the book define AGI?" }, "options": [ { "modality": "text", "value": "Any model trained with reinforcement learning on a closed game", "short": "RL-trained on a single closed game" }, { "modality": "text", "value": "A machine that predicts protein structures at experimental accuracy", "short": "Predicts protein structures precisely" }, { "modality": "text", "value": "A system that beats human experts at one narrow, well-defined task", "short": "Beats experts at one narrow task" }, { "modality": "text", "value": "A system matching or surpassing humans across the full range of cognitive tasks", "short": "Human-level across all cognitive tasks" } ], "correctIndex": 3, "explanation": "AGI spans the full cognitive range, not one narrow domain. Beating experts at a single task — Go, or protein folding — is exactly the narrow capability AGI is contrasted with.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1b5q7twrtusxu" }, { "id": "ot-agi-3", "shape": "mcq", "tags": [ "agi", "origin" ], "prompt": { "modality": "text", "value": "Who coined the term AGI, and in what year?" }, "options": [ { "modality": "text", "value": "Marcus Hutter, in 2002" }, { "modality": "text", "value": "Shane Legg, in 2002" }, { "modality": "text", "value": "Demis Hassabis, in 2010" }, { "modality": "text", "value": "Sebastian Mallaby, in 2026" } ], "correctIndex": 1, "explanation": "Shane Legg coined AGI in 2002, in conversations with Marcus Hutter; it was later adopted as DeepMind's founding mission. Hutter was in the conversation but did not coin it.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1g6onmml7r228" }, { "id": "ot-agi-4", "shape": "mcq", "tags": [ "agi", "apply" ], "prompt": { "modality": "text", "value": "Which system would Mallaby count as an infinity machine itself, rather than a step toward one?" }, "options": [ { "modality": "text", "value": "A system that masters a single board game well enough to beat the world's champions", "short": "Masters one board game vs champions" }, { "modality": "text", "value": "A system able, in principle, to solve any problem a human mind can frame", "short": "Solves any problem a human can frame" }, { "modality": "text", "value": "A system that learns to play many arcade games from raw screen pixels", "short": "Learns arcade games from raw pixels" }, { "modality": "text", "value": "A system that predicts a protein's three-dimensional structure from its amino-acid sequence", "short": "Predicts protein structure from sequence" } ], "correctIndex": 1, "explanation": "The infinity machine is the full cognitive range — any problem a human can frame. AlphaFold, AlphaGo and the Atari DQN are narrow milestones on the path, however impressive.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "3vu53wnkf0i2" }, { "id": "ot-agi-5", "shape": "mcq", "tags": [ "agi", "construction" ], "prompt": { "modality": "text", "value": "According to the book, how is the infinity machine actually being constructed?" }, "options": [ { "modality": "text", "value": "By combining existing narrow systems into a committee", "short": "A committee of narrow systems" }, { "modality": "text", "value": "Through successive, linked breakthroughs rather than one dramatic leap", "short": "Linked breakthroughs, not one leap" }, { "modality": "text", "value": "Through a single moonshot project inside Google", "short": "One moonshot project at Google" }, { "modality": "text", "value": "By scaling one narrow game-playing agent indefinitely", "short": "Scaling one game agent forever" } ], "correctIndex": 1, "explanation": "Mallaby's premise is that it comes from successive linked breakthroughs — reinforcement learning, deep learning, transformers — not a single dramatic leap or moonshot.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "d6joffzsk6qt" }, { "id": "ot-thesis-1", "shape": "mcq", "tags": [ "thesis", "argument" ], "prompt": { "modality": "text", "value": "What is Mallaby's one-line thesis about the century's most consequential technology?" }, "options": [ { "modality": "text", "value": "The technology's path is determined by compute budgets alone", "short": "Compute budgets alone decide it" }, { "modality": "text", "value": "Government regulation will decide what ultimately gets built", "short": "Regulation decides what's built" }, { "modality": "text", "value": "The builders' biographies and rivalries are determinative of the outcome", "short": "The builders themselves decide it" }, { "modality": "text", "value": "The outcome was fixed once the transformer was invented", "short": "The transformer already fixed it" } ], "correctIndex": 2, "explanation": "Mallaby argues the people's biographies and rivalries are not incidental but determinative. He never claims compute or regulation settles the outcome.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1gywox27umw6g" }, { "id": "ot-thesis-2", "shape": "mcq", "tags": [ "deepmind", "founding" ], "prompt": { "modality": "text", "value": "DeepMind was founded out of a London poker game and a pitch delivered at which event?" }, "options": [ { "modality": "text", "value": "A Google I/O keynote" }, { "modality": "text", "value": "The Singularity Summit" }, { "modality": "text", "value": "The Royal Society's AI symposium" }, { "modality": "text", "value": "The Turing Institute launch" } ], "correctIndex": 1, "explanation": "The book describes DeepMind as born of a London poker game and a pitch at the Singularity Summit.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "771ns51fuwl9f" }, { "id": "ot-thesis-3", "shape": "mcq", "tags": [ "thesis", "hassabis" ], "prompt": { "modality": "text", "value": "What does Mallaby say about Hassabis's view of slowing the AI race?" }, "options": [ { "modality": "text", "value": "Hassabis has withdrawn from the race on principle", "short": "He withdrew from the race" }, { "modality": "text", "value": "Hassabis does not believe any single actor can slow it; he is a participant", "short": "No single actor can slow it" }, { "modality": "text", "value": "Hassabis argues regulation has already halted the race", "short": "Regulation already halted it" }, { "modality": "text", "value": "Hassabis believes DeepMind alone can pause the race at will", "short": "DeepMind alone can pause it" } ], "correctIndex": 1, "explanation": "Mallaby is explicit: Hassabis doubts any single actor can slow the race down — he is a participant, not a brake.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "422e2n1ghj47x" }, { "id": "ot-thesis-4", "shape": "mcq", "tags": [ "thesis", "which-not" ], "prompt": { "modality": "text", "value": "Which is NOT part of Mallaby's thesis about DeepMind?" }, "options": [ { "modality": "text", "value": "Its people's rivalries shaped what got built" }, { "modality": "text", "value": "It framed AGI as a primary goal, not a product feature" }, { "modality": "text", "value": "It shaped an entire field's self-understanding" }, { "modality": "text", "value": "Its influence has been proportionate to its modest size" } ], "correctIndex": 3, "explanation": "Mallaby argues DeepMind's influence has been disproportionate to its size — the opposite of proportionate. The other three are core claims of the thesis.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "czc9396buakj" }, { "id": "ot-origins-1", "shape": "mcq", "tags": [ "hassabis", "biography" ], "prompt": { "modality": "text", "value": "In which year was Demis Hassabis born?" }, "options": [ { "modality": "text", "value": "1980" }, { "modality": "text", "value": "1984" }, { "modality": "text", "value": "1972" }, { "modality": "text", "value": "1976" } ], "correctIndex": 3, "explanation": "Hassabis was born in 1976. 1984 is a tempting misread — that is roughly when, at age eight, he bought his ZX Spectrum with his chess winnings.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "nwv5wg152nbf6" }, { "id": "ot-origins-2", "shape": "mcq", "tags": [ "hassabis", "family" ], "prompt": { "modality": "text", "value": "What was the heritage of Hassabis's mother?" }, "options": [ { "modality": "text", "value": "Greek Cypriot" }, { "modality": "text", "value": "Chinese Malaysian" }, { "modality": "text", "value": "Singaporean Indian" }, { "modality": "text", "value": "Chinese Singaporean" } ], "correctIndex": 3, "explanation": "His mother was Chinese Singaporean. Greek Cypriot is the classic swap — that was his father's heritage, not his mother's.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1qta8qysnnur4" }, { "id": "ot-origins-3", "shape": "mcq", "tags": [ "hassabis", "family" ], "prompt": { "modality": "text", "value": "What was the heritage of Hassabis's father?" }, "options": [ { "modality": "text", "value": "Greek mainland" }, { "modality": "text", "value": "Chinese Singaporean" }, { "modality": "text", "value": "Turkish Cypriot" }, { "modality": "text", "value": "Greek Cypriot" } ], "correctIndex": 3, "explanation": "His father was Greek Cypriot — specifically Cypriot, not mainland Greek. Chinese Singaporean was his mother's heritage.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1990aapksbd4f" }, { "id": "ot-origins-4", "shape": "mcq", "tags": [ "hassabis", "biography" ], "prompt": { "modality": "text", "value": "Which of these statements about Hassabis's origins is NOT accurate?" }, "options": [ { "modality": "text", "value": "He was born in Finchley, north London" }, { "modality": "text", "value": "The family lived above the shop his father ran" }, { "modality": "text", "value": "He was born in Guildford, Surrey" }, { "modality": "text", "value": "As a toddler he wanted to take apart every game in the shop" } ], "correctIndex": 2, "explanation": "He was born in Finchley, north London. Guildford is where Molyneux's Bullfrog studio was based — a place from later in the chapter, not his birthplace.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "bnbc5ly3xzun" }, { "id": "ot-origins-5", "shape": "mcq", "tags": [ "hassabis", "biography" ], "prompt": { "modality": "text", "value": "In which month was Demis Hassabis born?" }, "options": [ { "modality": "text", "value": "January" }, { "modality": "text", "value": "July" }, { "modality": "text", "value": "October" }, { "modality": "text", "value": "March" } ], "correctIndex": 1, "explanation": "Mallaby dates the birth precisely to July, in Finchley. The other months are plausible but the chapter names only July.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "113nw2o1fct6ja" }, { "id": "ot-chess-1", "shape": "mcq", "tags": [ "chess", "prodigy" ], "prompt": { "modality": "text", "value": "At what age did Hassabis reach chess master strength?" }, "options": [ { "modality": "text", "value": "11" }, { "modality": "text", "value": "15" }, { "modality": "text", "value": "17" }, { "modality": "text", "value": "13" } ], "correctIndex": 3, "explanation": "By age 13 Hassabis had reached chess master strength — Mallaby's marker of how far the prodigy had already gone before his teens were out.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "ep8ua1x2yczw" }, { "id": "ot-chess-2", "shape": "mcq", "tags": [ "chess", "deep-blue" ], "prompt": { "modality": "text", "value": "In which year did Deep Blue defeat Garry Kasparov?" }, "options": [ { "modality": "text", "value": "1997" }, { "modality": "text", "value": "1991" }, { "modality": "text", "value": "2001" }, { "modality": "text", "value": "1993" } ], "correctIndex": 0, "explanation": "Deep Blue beat Kasparov in 1997. 1991 is a tempting distractor — that is when the teenage Hassabis first experimented with PDP-style backpropagation.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "ucprnoihwkga" }, { "id": "ot-chess-3", "shape": "mcq", "tags": [ "chess", "prodigy" ], "prompt": { "modality": "text", "value": "At his peak as a junior player, what world ranking did Hassabis briefly reach for his age group?" }, "options": [ { "modality": "text", "value": "Tenth-highest-rated" }, { "modality": "text", "value": "Fifth-highest-rated" }, { "modality": "text", "value": "Second-highest-rated" }, { "modality": "text", "value": "The top-ranked (1st)" } ], "correctIndex": 2, "explanation": "He was briefly the second-highest-rated player in the world for his age group — remarkable, but not the outright top spot.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "d5iy1y1u4d998" }, { "id": "ot-chess-4", "shape": "mcq", "tags": [ "chess", "prodigy" ], "prompt": { "modality": "text", "value": "Which of these statements about Hassabis's chess years is NOT accurate?" }, "options": [ { "modality": "text", "value": "He picked up chess at four, watching his father play", "short": "Started at 4, watching his father" }, { "modality": "text", "value": "Within months of learning he was beating adults", "short": "Was beating adults within months" }, { "modality": "text", "value": "He took up chess at eight through a school chess club", "short": "Took up chess at 8 via a school club" }, { "modality": "text", "value": "He kept playing chess but lost interest in tournaments as a teenager", "short": "Kept playing but quit teen tournaments" } ], "correctIndex": 2, "explanation": "He picked up chess at four by watching his father, not at eight through a club — eight is the age he bought the ZX Spectrum. The other three statements are all in the chapter.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "125egmjihpna5" }, { "id": "ot-formative-1", "shape": "mcq", "tags": [ "zx-spectrum", "coding" ], "prompt": { "modality": "text", "value": "Which home computer did the eight-year-old Hassabis buy with his chess winnings?" }, "options": [ { "modality": "text", "value": "Commodore 64" }, { "modality": "text", "value": "Amstrad CPC 464" }, { "modality": "text", "value": "BBC Micro" }, { "modality": "text", "value": "ZX Spectrum" } ], "correctIndex": 3, "explanation": "He bought a ZX Spectrum and taught himself BASIC and machine code from the manuals. The other machines were contemporaries of the Spectrum but not the one he owned.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "8oo2dz1yf82o1" }, { "id": "ot-formative-2", "shape": "mcq", "tags": [ "geb", "influences" ], "prompt": { "modality": "text", "value": "Which book does Mallaby single out as the most-loved text of Hassabis's teenage years?" }, "options": [ { "modality": "text", "value": "The Selfish Gene" }, { "modality": "text", "value": "The Society of Mind" }, { "modality": "text", "value": "Gödel, Escher, Bach" }, { "modality": "text", "value": "The Emperor's New Mind" } ], "correctIndex": 2, "explanation": "Mallaby names Douglas Hofstadter's Gödel, Escher, Bach. The other titles are plausible period books about minds and machines, but the chapter singles out only GEB.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "15ezyy61921ii0" }, { "id": "ot-formative-3", "shape": "mcq", "tags": [ "zx-spectrum", "neural-networks" ], "prompt": { "modality": "text", "value": "Within a year of getting his home computer, what did Hassabis write that he did not yet have the theory to understand?" }, "options": [ { "modality": "text", "value": "A BASIC compiler" }, { "modality": "text", "value": "A chess engine that beat his father" }, { "modality": "text", "value": "A text adventure game" }, { "modality": "text", "value": "A primitive neural network" } ], "correctIndex": 3, "explanation": "By his own later account he coded a primitive neural network, simply wanting the computer to play games against itself — a straight line to AlphaZero thirty years on.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1um64r7d6ixrp" }, { "id": "ot-formative-4", "shape": "mcq", "tags": [ "geb", "intelligence" ], "prompt": { "modality": "text", "value": "What did Hofstadter's braided account of self-reference, formal systems and strange loops give Hassabis?" }, "options": [ { "modality": "text", "value": "A method for training neural networks with backpropagation", "short": "A method for training with backprop" }, { "modality": "text", "value": "A first sense that intelligence could be built from layered patterns rather than being mystical", "short": "Intelligence is built from patterns" }, { "modality": "text", "value": "A conviction that intelligence requires a biological brain", "short": "Intelligence needs a biological brain" }, { "modality": "text", "value": "A formal proof that machine intelligence is impossible", "short": "Proof machine minds are impossible" } ], "correctIndex": 1, "explanation": "GEB gave him his first articulate sense that intelligence might be built out of layered patterns rather than something mystical. Backprop came separately, via his 1991 brush with PDP.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "u5yj1qa9brw0" }, { "id": "ot-formative-5", "shape": "mcq", "tags": [ "zx-spectrum", "coding" ], "prompt": { "modality": "text", "value": "Which of these was NOT part of Hassabis's self-teaching on his home computer?" }, "options": [ { "modality": "text", "value": "Writing programs that made the computer play games against itself", "short": "Self-play game programs" }, { "modality": "text", "value": "Learning BASIC from manuals", "short": "BASIC from manuals" }, { "modality": "text", "value": "Learning machine code from manuals", "short": "Machine code from manuals" }, { "modality": "text", "value": "A structured programming course run by his school", "short": "A structured school course" } ], "correctIndex": 3, "explanation": "The chapter stresses he was self-taught from manuals — BASIC, machine code, and his own game-playing programs. No formal course is mentioned.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "ckhutn1rlmp1l" }, { "id": "ot-games-1", "shape": "mcq", "tags": [ "bullfrog", "games" ], "prompt": { "modality": "text", "value": "What was the teenage Hassabis's role on the Bullfrog management sim he worked on?" }, "options": [ { "modality": "text", "value": "Co-designer and lead programmer" }, { "modality": "text", "value": "Producer and marketing lead" }, { "modality": "text", "value": "Junior QA tester" }, { "modality": "text", "value": "Composer and sound designer" } ], "correctIndex": 0, "explanation": "He was co-designer and lead programmer on the management sim, which went on to sell millions — not a junior hire fetching bug reports.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "18zvp3j1ne73tt" }, { "id": "ot-games-2", "shape": "mcq", "tags": [ "molyneux", "mentors" ], "prompt": { "modality": "text", "value": "Who ran the studio that hired Hassabis and became one of his most important mentors?" }, "options": [ { "modality": "text", "value": "Peter Molyneux" }, { "modality": "text", "value": "Sid Meier" }, { "modality": "text", "value": "Richard Garriott" }, { "modality": "text", "value": "David Braben" } ], "correctIndex": 0, "explanation": "Peter Molyneux, already a celebrity inside game design, hired him and modelled the importance of audacious goals. The others are famous designers of the era but had no role here.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1ipo2pvuajex9" }, { "id": "ot-games-3", "shape": "mcq", "tags": [ "bullfrog", "timeline" ], "prompt": { "modality": "text", "value": "How old was Hassabis when Molyneux gave him a job at the studio?" }, "options": [ { "modality": "text", "value": "19" }, { "modality": "text", "value": "17" }, { "modality": "text", "value": "21" }, { "modality": "text", "value": "15" } ], "correctIndex": 1, "explanation": "Molyneux hired the seventeen-year-old during his gap year after A-levels, before Cambridge — so 19 and 21 are too late in his timeline.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "nydjhmjckjvs" }, { "id": "ot-games-4", "shape": "mcq", "tags": [ "mallaby", "games", "argument" ], "prompt": { "modality": "text", "value": "How does Mallaby's reading of Hassabis's games years differ from the standard biography?" }, "options": [ { "modality": "text", "value": "The standard version overstates Molyneux's influence; Mallaby credits Cambridge instead", "short": "Standard: Molyneux's role; Mallaby: Cambridge" }, { "modality": "text", "value": "The standard version dates the Bullfrog job too early; Mallaby moves it after his PhD", "short": "Standard: Bullfrog early; Mallaby: after PhD" }, { "modality": "text", "value": "The standard version says he left games for good after his gap year; Mallaby says he returned to Bullfrog later", "short": "Standard: quit for good; Mallaby: returned" }, { "modality": "text", "value": "The standard version treats games as a colourful prelude; Mallaby argues games were the real career", "short": "Standard: a prelude; Mallaby: his career" } ], "correctIndex": 3, "explanation": "Mallaby punctures the prelude framing: in his reading the neuroscience PhD and DeepMind were what Hassabis did to build better games — first virtual, then games played against nature.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "snewwm18s48pg" }, { "id": "ot-cambridge-1", "shape": "mcq", "tags": [ "cambridge", "david-silver" ], "prompt": { "modality": "text", "value": "Which fellow fresher did Hassabis meet in the Junior Common Room at Queens', beginning the longest professional partnership of his career?" }, "options": [ { "modality": "text", "value": "Peter Molyneux" }, { "modality": "text", "value": "Mustafa Suleyman" }, { "modality": "text", "value": "David Silver" }, { "modality": "text", "value": "Shane Legg" } ], "correctIndex": 2, "explanation": "Hassabis met David Silver in his first weeks at Queens'. Legg and Suleyman were later DeepMind co-founders, not Cambridge freshers; Molyneux was his boss at Bullfrog, before university.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "5vd7qcmlu23u" }, { "id": "ot-cambridge-2", "shape": "mcq", "tags": [ "cambridge", "go" ], "prompt": { "modality": "text", "value": "Hassabis and David Silver bonded over board games as Cambridge freshers. Which single game does the guide name in particular?" }, "options": [ { "modality": "text", "value": "Backgammon" }, { "modality": "text", "value": "Chess" }, { "modality": "text", "value": "Diplomacy" }, { "modality": "text", "value": "Go" } ], "correctIndex": 3, "explanation": "The guide says they bonded over board games — Go in particular. Go is the specific game named for the Cambridge friendship, and the one Silver would later attack with AlphaGo.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1llkik09s7t8i" }, { "id": "ot-cambridge-3", "shape": "mcq", "tags": [ "books", "hofstadter" ], "prompt": { "modality": "text", "value": "Mallaby singles out which book as the single most important one in Hassabis's intellectual formation?" }, "options": [ { "modality": "text", "value": "The Selfish Gene" }, { "modality": "text", "value": "Gödel, Escher, Bach" }, { "modality": "text", "value": "A Brief History of Time" }, { "modality": "text", "value": "The Society of Mind" } ], "correctIndex": 1, "explanation": "Douglas Hofstadter's Gödel, Escher, Bach is the book Mallaby singles out. Minsky's The Society of Mind is a tempting AI-adjacent choice but is not the one named.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "iz4gfpap8v0f" }, { "id": "ot-cambridge-4", "shape": "mcq", "tags": [ "david-silver", "alphago" ], "prompt": { "modality": "text", "value": "Which of the following is NOT true of David Silver?" }, "options": [ { "modality": "text", "value": "He was a Bullfrog colleague Hassabis recruited to Cambridge", "short": "Bullfrog colleague at Cambridge" }, { "modality": "text", "value": "He later contributed to the reinforcement-learning work inside AlphaFold", "short": "Worked on RL in AlphaFold" }, { "modality": "text", "value": "He met Hassabis in the Junior Common Room at Queens'", "short": "Met Hassabis in Queens' JCR" }, { "modality": "text", "value": "He would go on to lead the AlphaGo project at DeepMind", "short": "Led AlphaGo at DeepMind" } ], "correctIndex": 0, "explanation": "Silver was a fellow fresher Hassabis met at Queens', not a Bullfrog colleague he brought with him. The other three statements are all accurate.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1nsd72kr1adjm" }, { "id": "ot-cambridge-5", "shape": "mcq", "tags": [ "cambridge", "philosophy" ], "prompt": { "modality": "text", "value": "Hassabis and Silver shared a frustration that the CS syllabus sidestepped what Hassabis called what?" }, "options": [ { "modality": "text", "value": "The deep philosophical questions" }, { "modality": "text", "value": "The limits of symbolic AI" }, { "modality": "text", "value": "The hard problem of consciousness" }, { "modality": "text", "value": "The grand-challenge problems" } ], "correctIndex": 0, "explanation": "Hassabis's phrase was \"the deep philosophical questions\" — what is intelligence, and what would it mean to construct it. \"Grand-challenge problems\" is DeepMind-era vocabulary, not the Cambridge phrase.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1ma55bc1sf0yva" }, { "id": "ot-neuro-1", "shape": "mcq", "tags": [ "ucl", "maguire" ], "prompt": { "modality": "text", "value": "Which of the following is NOT true of Hassabis's UCL doctoral work?" }, "options": [ { "modality": "text", "value": "It was supervised by Eleanor Maguire", "short": "Supervised by Eleanor Maguire" }, { "modality": "text", "value": "He carried it out in the years before he founded Elixir Studios", "short": "Done before he founded Elixir" }, { "modality": "text", "value": "It studied patients with damage to the hippocampus", "short": "Studied hippocampal damage" }, { "modality": "text", "value": "It convinced him the brain is a generative model", "short": "Brain as a generative model" } ], "correctIndex": 1, "explanation": "The PhD came AFTER Elixir wound down in 2005 — he enrolled almost immediately afterwards. The other three statements are accurate.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1i57kk04n5w86" }, { "id": "ot-neuro-2", "shape": "mcq", "tags": [ "hippocampus", "neuroscience" ], "prompt": { "modality": "text", "value": "Damage to which brain region left Hassabis's patients unable to imagine future scenes?" }, "options": [ { "modality": "text", "value": "The amygdala" }, { "modality": "text", "value": "The cerebellum" }, { "modality": "text", "value": "The prefrontal cortex" }, { "modality": "text", "value": "The hippocampus" } ], "correctIndex": 3, "explanation": "The hippocampus — already known for storing memories — turned out to also construct hypothetical futures. The amygdala handles emotional processing and is not the region in the finding.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1w7vlxcu5b3ce" }, { "id": "ot-neuro-3", "shape": "mcq", "tags": [ "memory", "imagination" ], "prompt": { "modality": "text", "value": "What did Hassabis's finding establish about the brain region that stores memories?" }, "options": [ { "modality": "text", "value": "It stores memories as exact, retrievable copies" }, { "modality": "text", "value": "It is separate from the region handling imagination" }, { "modality": "text", "value": "It only becomes active during sleep consolidation" }, { "modality": "text", "value": "It is also responsible for constructing hypothetical futures" } ], "correctIndex": 3, "explanation": "The point of the finding is that one region does both jobs — storing memories and constructing imagined futures. The idea that memory and imagination live in separate regions is exactly what the result overturned.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "zo9xrg1m4jffe" }, { "id": "ot-neuro-4", "shape": "mcq", "tags": [ "hippocampus", "generative-model" ], "prompt": { "modality": "text", "value": "A patient with hippocampal damage is asked to describe lying on a beach they have never visited. Per Hassabis's finding, what happens?" }, "options": [ { "modality": "text", "value": "They cannot construct the scene at all" }, { "modality": "text", "value": "They construct it vividly but get the details wrong" }, { "modality": "text", "value": "They can construct it, but cannot recall any real past beach" }, { "modality": "text", "value": "They construct it only after being shown a photograph" } ], "correctIndex": 0, "explanation": "Hassabis's patients could not imagine future or hypothetical scenes — the construction itself fails. Option 3 inverts the finding: their deficit is in building novel scenes, not only in retrieval.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1mlukyo1qkcqvm" }, { "id": "ot-elixir-founding-1", "shape": "mcq", "tags": [ "elixir", "founding" ], "prompt": { "modality": "text", "value": "On what date did Hassabis found Elixir Studios?" }, "options": [ { "modality": "text", "value": "5 May 2003" }, { "modality": "text", "value": "1 January 2005" }, { "modality": "text", "value": "7 July 1998" }, { "modality": "text", "value": "7 July 1994" } ], "correctIndex": 2, "explanation": "Elixir Studios was founded on 7 July 1998. 1994 is when Hassabis went up to Cambridge, 2003 is Republic's release, and 2005 is when Elixir wound down.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "876cq1gnntfz" }, { "id": "ot-elixir-founding-2", "shape": "mcq", "tags": [ "elixir", "founding" ], "prompt": { "modality": "text", "value": "How old was Hassabis when he founded Elixir Studios?" }, "options": [ { "modality": "text", "value": "26" }, { "modality": "text", "value": "30" }, { "modality": "text", "value": "22" }, { "modality": "text", "value": "17" } ], "correctIndex": 2, "explanation": "He was 22 when he founded Elixir on 7 July 1998, a year after graduating Cambridge in 1997 — which also rules out the older ages.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1kdi7nznf2hal" }, { "id": "ot-elixir-founding-3", "shape": "mcq", "tags": [ "elixir", "definition" ], "prompt": { "modality": "text", "value": "Which best defines Elixir Studios?" }, "options": [ { "modality": "text", "value": "The Cambridge research group where Hassabis and Silver studied Go", "short": "Cambridge Go research group" }, { "modality": "text", "value": "The UCL neuroscience lab Hassabis joined for his PhD", "short": "UCL neuroscience PhD lab" }, { "modality": "text", "value": "The publisher that funded and distributed Bullfrog's games", "short": "Bullfrog's games publisher" }, { "modality": "text", "value": "The games company Hassabis founded after leaving Bullfrog, building ambitious AI-driven games", "short": "Hassabis's AI games company" } ], "correctIndex": 3, "explanation": "Elixir Studios was Hassabis's own games company, founded post-Bullfrog to build ambitious AI-driven games; it struggled commercially and closed in 2005. It was never a lab, a research group, or a publisher.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1sgb8c5tg9kif" }, { "id": "ot-elixir-founding-4", "shape": "mcq", "tags": [ "elixir", "london", "bullfrog" ], "prompt": { "modality": "text", "value": "Where did Hassabis set up Elixir Studios, and what premise did he build it on?" }, "options": [ { "modality": "text", "value": "Guildford, on the premise of licensing other studios' AI engines", "short": "Guildford; licensing others' AI engines" }, { "modality": "text", "value": "Central London, on the premise of low-cost casual games at high volume", "short": "Central London; low-cost casual games" }, { "modality": "text", "value": "Cambridge, on the premise of commercialising academic neuroscience", "short": "Cambridge; commercialising neuroscience" }, { "modality": "text", "value": "Central London, on Bullfrog's premise of ambitious AI baked into games", "short": "Central London; ambitious AI in games" } ], "correctIndex": 3, "explanation": "Elixir was founded in central London on the premise Bullfrog had taught him: ambitious AI baked into games at a level no other studio would attempt — the opposite of a low-cost casual strategy.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "11fq6v15cnc6r" }, { "id": "ot-elixir-founding-5", "shape": "mcq", "tags": [ "elixir", "end" ], "prompt": { "modality": "text", "value": "Which of the following is NOT true of Elixir Studios?" }, "options": [ { "modality": "text", "value": "It was wound down by 2005" }, { "modality": "text", "value": "It built games prized for their AI ambition" }, { "modality": "text", "value": "It was founded in 1998 when Hassabis was 22" }, { "modality": "text", "value": "It was a commercial success that Hassabis sold at a profit" } ], "correctIndex": 3, "explanation": "Elixir failed commercially; by 2005 it was wound down and Hassabis merely sold the rights to its completed games and walked away. The other three statements are accurate.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1tjx4nrlrm5wd" }, { "id": "ot-elixir-games-1", "shape": "mcq", "tags": [ "evil-genius", "elixir" ], "prompt": { "modality": "text", "value": "Which Elixir game was a satirical spy-villain strategy game released in 2004?" }, "options": [ { "modality": "text", "value": "Dungeon Keeper" }, { "modality": "text", "value": "No One Lives Forever" }, { "modality": "text", "value": "Theme Park" }, { "modality": "text", "value": "Evil Genius" } ], "correctIndex": 3, "explanation": "Evil Genius (2004) was Elixir's satirical spy-villain strategy game. Theme Park and Dungeon Keeper are Bullfrog titles, not Elixir's.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "y69krvesnmwd" }, { "id": "ot-elixir-games-2", "shape": "mcq", "tags": [ "republic", "elixir" ], "prompt": { "modality": "text", "value": "Republic: The Revolution was set against what backdrop?" }, "options": [ { "modality": "text", "value": "A South American drug cartel war" }, { "modality": "text", "value": "A fictional Eastern European revolution" }, { "modality": "text", "value": "A Cold War espionage network in Berlin" }, { "modality": "text", "value": "A near-future Martian colony uprising" } ], "correctIndex": 1, "explanation": "Republic modelled thousands of citizens with individual goals, opinions, and relationships against a fictional Eastern European revolution. The espionage setting belongs to Evil Genius's genre, not Republic's.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "swl4hn17snfg1" }, { "id": "ot-elixir-games-3", "shape": "mcq", "tags": [ "elixir", "reception" ], "prompt": { "modality": "text", "value": "How did Elixir's two best-known games perform?" }, "options": [ { "modality": "text", "value": "Both were commercial smashes but panned for weak AI", "short": "Commercial hits, panned for weak AI" }, { "modality": "text", "value": "Republic sold well while Evil Genius was never finished", "short": "Republic sold; Evil Genius unfinished" }, { "modality": "text", "value": "Both were cancelled before release when Elixir closed", "short": "Both cancelled before release" }, { "modality": "text", "value": "Both were prized for AI ambition, but neither was a commercial smash", "short": "Praised for AI, not commercial hits" } ], "correctIndex": 3, "explanation": "Republic and Evil Genius were both prized for AI ambition and neither was a commercial smash — Evil Genius performed better but still could not rescue the studio's finances. Both did ship.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "mv91qsca6sni" }, { "id": "ot-elixir-games-4", "shape": "mcq", "tags": [ "republic", "hardware" ], "prompt": { "modality": "text", "value": "Critics admired Republic: The Revolution's concept but found the gameplay sluggish. Why?" }, "options": [ { "modality": "text", "value": "Early-2000s hardware could not support what Hassabis asked of it", "short": "Early-2000s hardware too weak" }, { "modality": "text", "value": "The game modelled only a dozen citizens, too few to feel alive", "short": "Modelled too few citizens" }, { "modality": "text", "value": "The studio shipped it a year before the AI was written", "short": "Shipped before the AI existed" }, { "modality": "text", "value": "Its publisher forced a rewrite into a different genre", "short": "Publisher forced a genre rewrite" } ], "correctIndex": 0, "explanation": "The concept overshot the era's compute — hardware timing, one of Mallaby's three Elixir lessons. The AI did exist and modelled thousands of citizens; it was simply too heavy for 2003 machines.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1k6akpyhk34zc" }, { "id": "ot-legg-agi-1", "shape": "mcq", "tags": [ "agi", "shane-legg" ], "prompt": { "modality": "text", "value": "Who proposed the phrase that became the field's agreed name for its most ambitious goal?" }, "options": [ { "modality": "text", "value": "Marcus Hutter" }, { "modality": "text", "value": "Shane Legg" }, { "modality": "text", "value": "Jürgen Schmidhuber" }, { "modality": "text", "value": "Demis Hassabis" } ], "correctIndex": 1, "explanation": "Legg proposed \"Artificial General Intelligence\" while brainstorming a book title with Hutter. Hutter was in the room but did not propose the phrase; Schmidhuber worked with Legg on recurrent nets, not the term.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "4a8nsch0rs9m" }, { "id": "ot-legg-agi-2", "shape": "mcq", "tags": [ "agi", "dates" ], "prompt": { "modality": "text", "value": "Around what year did Shane Legg propose the term \"Artificial General Intelligence\"?" }, "options": [ { "modality": "text", "value": "2002" }, { "modality": "text", "value": "2010" }, { "modality": "text", "value": "2009" }, { "modality": "text", "value": "1997" } ], "correctIndex": 0, "explanation": "The guide dates the coinage to around 2002, during the book-title brainstorm with Hutter. 2009 is the Halloween Scenario lecture; 2010 is the poker night and the Thiel pitch.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "o4ad2h2ekfzz" }, { "id": "ot-legg-agi-3", "shape": "mcq", "tags": [ "idsia", "schmidhuber" ], "prompt": { "modality": "text", "value": "At IDSIA, which collaborator did Legg work with on recurrent neural networks?" }, "options": [ { "modality": "text", "value": "David Silver" }, { "modality": "text", "value": "Geoffrey Hinton" }, { "modality": "text", "value": "Jürgen Schmidhuber" }, { "modality": "text", "value": "Marcus Hutter" } ], "correctIndex": 2, "explanation": "Legg worked with Schmidhuber on recurrent neural networks and with Hutter on theories of universal intelligence — Hutter is the tempting swap. Hinton and Silver are not part of Legg's IDSIA years.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1p54clv1l4g1x9" }, { "id": "ot-legg-agi-4", "shape": "mcq", "tags": [ "agi", "shane-legg" ], "prompt": { "modality": "text", "value": "By the time Hassabis met Legg, what standing did the phrase Legg had proposed already have?" }, "options": [ { "modality": "text", "value": "It was the agreed name for the field's most ambitious goal" }, { "modality": "text", "value": "It was still a private joke between Legg and Hutter" }, { "modality": "text", "value": "It had not yet been proposed" }, { "modality": "text", "value": "It had been dropped in favour of \"universal intelligence\"" } ], "correctIndex": 0, "explanation": "The guide says the phrase stuck: by the time Hassabis met Legg, \"Artificial General Intelligence\" was the agreed name for the field's most ambitious goal. Universal intelligence was Legg and Hutter's IDSIA research topic, not a rival name.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "12gsffja73br9" }, { "id": "ot-idsia-1", "shape": "mcq", "tags": [ "idsia" ], "prompt": { "modality": "text", "value": "Which Swiss AI institute did Shane Legg land at, working with Marcus Hutter and Jürgen Schmidhuber?" }, "options": [ { "modality": "text", "value": "ETH Zurich" }, { "modality": "text", "value": "Idiap Research Institute" }, { "modality": "text", "value": "IDSIA" }, { "modality": "text", "value": "EPFL" } ], "correctIndex": 2, "explanation": "IDSIA is the Swiss AI institute where Legg worked with Hutter and Schmidhuber. EPFL, ETH Zurich and Idiap are real Swiss research institutions, but none is the lab named in the guide.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1s3tn91hlf3w7" }, { "id": "ot-idsia-2", "shape": "mcq", "tags": [ "idsia", "geography" ], "prompt": { "modality": "text", "value": "In which city is IDSIA located?" }, "options": [ { "modality": "text", "value": "Geneva" }, { "modality": "text", "value": "Lugano" }, { "modality": "text", "value": "Zurich" }, { "modality": "text", "value": "Lausanne" } ], "correctIndex": 1, "explanation": "The guide places IDSIA, the Swiss AI institute, in Lugano. Zurich, Geneva and Lausanne are other Swiss research cities but not IDSIA's home.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "x0k9bm1itxea4" }, { "id": "ot-idsia-3", "shape": "mcq", "tags": [ "idsia", "ucl" ], "prompt": { "modality": "text", "value": "Which of these is NOT connected to IDSIA in the guide's account?" }, "options": [ { "modality": "text", "value": "Shane Legg" }, { "modality": "text", "value": "Jürgen Schmidhuber" }, { "modality": "text", "value": "Demis Hassabis" }, { "modality": "text", "value": "Marcus Hutter" } ], "correctIndex": 2, "explanation": "Shane Legg, Jürgen Schmidhuber, and Marcus Hutter were all researchers at IDSIA in Lugano — Schmidhuber as director, Legg and Hutter as students. Demis Hassabis, co-founder of DeepMind alongside Legg, came up through a different path (games and UCL neuroscience), with no IDSIA connection.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "8zes6s191frf2" }, { "id": "ot-idsia-4", "shape": "mcq", "tags": [ "idsia", "universal-intelligence" ], "prompt": { "modality": "text", "value": "What did Legg and Hutter work on together at IDSIA?" }, "options": [ { "modality": "text", "value": "Theories of universal intelligence" }, { "modality": "text", "value": "Climate-diplomacy modelling" }, { "modality": "text", "value": "Protein-structure prediction" }, { "modality": "text", "value": "Reinforcement learning for Atari games" } ], "correctIndex": 0, "explanation": "At IDSIA, Legg worked with Hutter on theories of universal intelligence; the recurrent-neural-network work was with Schmidhuber. Atari and protein folding are later DeepMind work, and climate diplomacy was Suleyman's consulting.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "15h7j8fqjwpsh" }, { "id": "ot-idsia-5", "shape": "mcq", "tags": [ "idsia", "shane-legg" ], "prompt": { "modality": "text", "value": "Before landing at IDSIA, Shane Legg had drifted through which fields?" }, "options": [ { "modality": "text", "value": "Cognitive science, finance and machine learning" }, { "modality": "text", "value": "Conflict resolution and climate diplomacy" }, { "modality": "text", "value": "Neuroscience, chess and video-game design" }, { "modality": "text", "value": "Law, journalism and public policy" } ], "correctIndex": 0, "explanation": "The guide says Legg drifted through cognitive science, finance and machine learning before IDSIA. Games and neuroscience describe Hassabis; conflict resolution and climate diplomacy describe Suleyman.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1jpnf65wr2fpn" }, { "id": "ot-suleyman-1", "shape": "mcq", "tags": [ "suleyman", "nickname" ], "prompt": { "modality": "text", "value": "What was Mustafa Suleyman's nickname inside DeepMind?" }, "options": [ { "modality": "text", "value": "Mus" }, { "modality": "text", "value": "Mo" }, { "modality": "text", "value": "Moose" }, { "modality": "text", "value": "Sulley" } ], "correctIndex": 2, "explanation": "Mallaby records the nickname as \"Moose\" — Mustafa \"Moose\" Suleyman. The other options are plausible shortenings of his name but are not the nickname used.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "q3mfvy1jwy6f8" }, { "id": "ot-suleyman-2", "shape": "mcq", "tags": [ "suleyman", "dates" ], "prompt": { "modality": "text", "value": "In what year was Mustafa Suleyman born?" }, "options": [ { "modality": "text", "value": "1973" }, { "modality": "text", "value": "1980" }, { "modality": "text", "value": "1990" }, { "modality": "text", "value": "1984" } ], "correctIndex": 3, "explanation": "Suleyman was born in August 1984 in north London. 1973 is Shane Legg's birth year — the tempting mix-up between the two cofounders.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1lbir29a9r503" }, { "id": "ot-suleyman-3", "shape": "mcq", "tags": [ "suleyman", "biography" ], "prompt": { "modality": "text", "value": "Which statement about Suleyman before DeepMind is NOT true?" }, "options": [ { "modality": "text", "value": "He dropped out of Oxford" }, { "modality": "text", "value": "He started a Muslim youth helpline" }, { "modality": "text", "value": "He held a doctorate in machine learning" }, { "modality": "text", "value": "He was a conflict-resolution consultant in Copenhagen" } ], "correctIndex": 2, "explanation": "Suleyman had no technical training in AI at all — he was the operator, not the researcher. The Oxford dropout, the youth helpline and the Copenhagen consulting work are all in the guide.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "14pycbl13mhfpf" }, { "id": "ot-suleyman-4", "shape": "mcq", "tags": [ "suleyman", "cofounders" ], "prompt": { "modality": "text", "value": "In Mallaby's triangle of the three cofounders, how is the third cofounder characterised?" }, "options": [ { "modality": "text", "value": "The relentless operator who could pitch a billionaire at midnight", "short": "The relentless operator-pitcher" }, { "modality": "text", "value": "The cerebral, safety-aware theorist", "short": "The cerebral safety theorist" }, { "modality": "text", "value": "The obsessive games-and-neuroscience polymath", "short": "The games-neuroscience polymath" }, { "modality": "text", "value": "The recurrent-neural-network specialist from Lugano", "short": "The RNN specialist from Lugano" } ], "correctIndex": 0, "explanation": "Suleyman is the operator of the triangle. The cerebral safety theorist is Legg; the games-and-neuroscience polymath is Hassabis; the Lugano recurrent-net specialist is Schmidhuber, who was never a cofounder.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "pgvlapa389j" }, { "id": "ot-connections-1", "shape": "mcq", "tags": [ "george-hassabis", "founding" ], "prompt": { "modality": "text", "value": "Through whom did Demis Hassabis first meet Mustafa Suleyman?" }, "options": [ { "modality": "text", "value": "Shane Legg" }, { "modality": "text", "value": "Luke Nosek" }, { "modality": "text", "value": "Peter Thiel" }, { "modality": "text", "value": "George Hassabis" } ], "correctIndex": 3, "explanation": "Demis's brother George had been at school with Suleyman and made the introduction. Nosek and Thiel enter the story later, at the 2010 fundraise; Legg met Hassabis separately at UCL.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1b6bo8cn7y2ha" }, { "id": "ot-connections-2", "shape": "mcq", "tags": [ "george-hassabis", "founding" ], "prompt": { "modality": "text", "value": "What was George Hassabis's relationship to the two men he connected?" }, "options": [ { "modality": "text", "value": "Demis's business partner, who met Suleyman at Oxford" }, { "modality": "text", "value": "Demis's brother, who met Suleyman at the Vic Casino" }, { "modality": "text", "value": "Demis's cousin, who worked with Suleyman in Copenhagen" }, { "modality": "text", "value": "Demis's brother, who had been at school with Suleyman" } ], "correctIndex": 3, "explanation": "George was Demis's brother and had been at school with Suleyman — that school tie is the hinge. The Vic Casino was the July 2010 poker night among the three cofounders, not where George met Suleyman.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1c6sjjzmxg3r9" }, { "id": "ot-connections-3", "shape": "mcq", "tags": [ "luke-nosek", "thiel" ], "prompt": { "modality": "text", "value": "Peter Thiel muttered \"Demis's destiny\" to which person?" }, "options": [ { "modality": "text", "value": "George Hassabis" }, { "modality": "text", "value": "Mustafa Suleyman" }, { "modality": "text", "value": "Shane Legg" }, { "modality": "text", "value": "Luke Nosek" } ], "correctIndex": 3, "explanation": "Thiel made the remark to his Founders Fund partner Luke Nosek as they walked the three founders back to their car. The cofounders were the subject of the remark, not its audience.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "z45mqp1pridxb" }, { "id": "ot-connections-4", "shape": "mcq", "tags": [ "luke-nosek", "founders-fund" ], "prompt": { "modality": "text", "value": "Who was Luke Nosek in this episode?" }, "options": [ { "modality": "text", "value": "DeepMind's first employee" }, { "modality": "text", "value": "Peter Thiel's partner at Founders Fund" }, { "modality": "text", "value": "Suleyman's colleague from his Copenhagen consulting work" }, { "modality": "text", "value": "An organiser of the Singularity Summit" } ], "correctIndex": 1, "explanation": "Nosek was Thiel's Founders Fund partner, present as the founders were walked back to their car. Founders Fund committed at the Palo Alto breakfast — but Nosek's role was investor-side, not a DeepMind hire.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1rgkcbi198fnbs" }, { "id": "ot-connections-5", "shape": "mcq", "tags": [ "thiel", "luke-nosek" ], "prompt": { "modality": "text", "value": "In what setting was the \"Demis's destiny\" remark made?" }, "options": [ { "modality": "text", "value": "On stage at the Singularity Summit in San Francisco", "short": "On stage at the Singularity Summit" }, { "modality": "text", "value": "Over poker at the Vic Casino in west London", "short": "Over poker at the Vic Casino" }, { "modality": "text", "value": "During Legg's \"Halloween Scenario\" lecture at UCL", "short": "At Legg's Halloween lecture, UCL" }, { "modality": "text", "value": "Walking the founders back to their car after breakfast at Thiel's Palo Alto house", "short": "After breakfast at Thiel's house" } ], "correctIndex": 3, "explanation": "Thiel muttered it to Luke Nosek while walking the trio back to their car after the Palo Alto breakfast. The Summit was where Suleyman engineered the introduction the previous day; the Vic poker night was a month earlier.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1pgc36lt0srxb" }, { "id": "ot-naming-1", "shape": "mcq", "tags": [ "naming", "founding" ], "prompt": { "modality": "text", "value": "What was DeepMind's first working name, before the founders settled on the name they kept?" }, "options": [ { "modality": "text", "value": "Foundation" }, { "modality": "text", "value": "Aurora" }, { "modality": "text", "value": "Solaria" }, { "modality": "text", "value": "Trantor" } ], "correctIndex": 2, "explanation": "The first working name was Solaria, borrowed from an Isaac Asimov novel. Trantor and Foundation also come from Asimov's fiction, which makes them tempting — but Solaria is the name the founders actually used.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "18viffh8iq41f" }, { "id": "ot-naming-2", "shape": "mcq", "tags": [ "naming", "asimov" ], "prompt": { "modality": "text", "value": "The company's original working name was borrowed from a novel by which author?" }, "options": [ { "modality": "text", "value": "Arthur C. Clarke" }, { "modality": "text", "value": "Isaac Asimov" }, { "modality": "text", "value": "Philip K. Dick" }, { "modality": "text", "value": "William Gibson" } ], "correctIndex": 1, "explanation": "Solaria comes from an Isaac Asimov novel. Clarke, Dick and Gibson are all plausible science-fiction sources, but the guide names Asimov specifically.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1bd6w5h560c5j" }, { "id": "ot-naming-3", "shape": "mcq", "tags": [ "founding-year", "timeline" ], "prompt": { "modality": "text", "value": "In what year was the DeepMind name registered?" }, "options": [ { "modality": "text", "value": "2014" }, { "modality": "text", "value": "2012" }, { "modality": "text", "value": "2010" }, { "modality": "text", "value": "2011" } ], "correctIndex": 2, "explanation": "The name was registered in late 2010; the London company followed in 2011. 2012 is when Page and Brin began watching the lab, and 2014 is the Google acquisition.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "wg0pt07uv6iy" }, { "id": "ot-naming-4", "shape": "mcq", "tags": [ "naming", "founders" ], "prompt": { "modality": "text", "value": "Which founder pushed for a plainer name than the original working title?" }, "options": [ { "modality": "text", "value": "Shane Legg" }, { "modality": "text", "value": "Mustafa Suleyman" }, { "modality": "text", "value": "Vlad Mnih" }, { "modality": "text", "value": "Demis Hassabis" } ], "correctIndex": 0, "explanation": "Shane Legg argued for something simpler than Solaria. Hassabis and Suleyman are cofounders too, but the guide credits Legg; Mnih was a researcher who later led the Atari DQN work.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1b0az2vpzwa7l" }, { "id": "ot-naming-5", "shape": "mcq", "tags": [ "naming", "deep-learning" ], "prompt": { "modality": "text", "value": "In the final name, what does the word \"Deep\" refer to?" }, "options": [ { "modality": "text", "value": "Deep Q-Networks, the architecture they had already built", "short": "Deep Q-Networks they built" }, { "modality": "text", "value": "The depth of the company's funding reserves", "short": "The depth of funding reserves" }, { "modality": "text", "value": "Deep learning, the neural-network approach they were betting on", "short": "Deep learning (neural nets)" }, { "modality": "text", "value": "Deep Blue, the chess computer that beat Kasparov", "short": "Deep Blue, the chess computer" } ], "correctIndex": 2, "explanation": "\"Deep\" nodded to deep learning; \"Mind\" declared the destination. DQN is tempting but came years later — the name was registered in 2010, before they had any product.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "9rb71h1el4qxb" }, { "id": "ot-acq-1", "shape": "mcq", "tags": [ "acquisition", "timeline" ], "prompt": { "modality": "text", "value": "In which month and year did Google close its acquisition of DeepMind?" }, "options": [ { "modality": "text", "value": "July 2013" }, { "modality": "text", "value": "October 2012" }, { "modality": "text", "value": "January 2011" }, { "modality": "text", "value": "January 2014" } ], "correctIndex": 3, "explanation": "Google closed the deal in January 2014. October 2012 is the Musk–Page dinner and July 2013 is the day the DQN agent learned Pong — both real dates in this lesson, but neither is the acquisition.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1d6m30vrbtr2d" }, { "id": "ot-acq-2", "shape": "mcq", "tags": [ "acquisition", "price" ], "prompt": { "modality": "text", "value": "Roughly how much did Google pay for DeepMind?" }, "options": [ { "modality": "text", "value": "£400 million" }, { "modality": "text", "value": "£4 billion" }, { "modality": "text", "value": "£40 million" }, { "modality": "text", "value": "£1.2 billion" } ], "correctIndex": 0, "explanation": "The price was roughly £400 million (about $600 million) — then the largest AI acquisition in European history. The other figures are off by an order of magnitude or more.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "mmyp111rzuse3" }, { "id": "ot-acq-3", "shape": "mcq", "tags": [ "acquisition", "price" ], "prompt": { "modality": "text", "value": "Google's purchase of DeepMind is often quoted in US dollars instead. Roughly what figure?" }, "options": [ { "modality": "text", "value": "$400 million" }, { "modality": "text", "value": "$1 billion" }, { "modality": "text", "value": "$600 million" }, { "modality": "text", "value": "$150 million" } ], "correctIndex": 2, "explanation": "Roughly $600 million, equivalent to about £400 million. $400 million is the pound figure with the wrong currency symbol — a common mix-up.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "xvsagt1nittsr" }, { "id": "ot-acq-4", "shape": "mcq", "tags": [ "acquisition", "larry-page" ], "prompt": { "modality": "text", "value": "Which pair of Google leaders had been watching DeepMind since 2012, ahead of the deal?" }, "options": [ { "modality": "text", "value": "Larry Page and Sergey Brin" }, { "modality": "text", "value": "Larry Page and Eric Schmidt" }, { "modality": "text", "value": "Eric Schmidt and Sundar Pichai" }, { "modality": "text", "value": "Sergey Brin and Sundar Pichai" } ], "correctIndex": 0, "explanation": "The guide names Google's cofounders Larry Page and Sergey Brin as the ones circling DeepMind from 2012. Schmidt and Pichai were Google executives but are not credited here.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1a1tk2bjrfmft" }, { "id": "ot-acq-5", "shape": "mcq", "tags": [ "acquisition", "ethics-board" ], "prompt": { "modality": "text", "value": "Which statement about the DeepMind acquisition is NOT true?" }, "options": [ { "modality": "text", "value": "The lab had fewer than 100 employees", "short": "Had fewer than 100 employees" }, { "modality": "text", "value": "Hassabis demanded an independent ethics board as a condition", "short": "Demanded an ethics board" }, { "modality": "text", "value": "DeepMind was already profitable on strong revenue at the time of the deal", "short": "Already profitable at the deal" }, { "modality": "text", "value": "It was then the largest AI acquisition in European history", "short": "Largest AI deal in Europe then" } ], "correctIndex": 2, "explanation": "DeepMind had no revenue at all when Google bought it — that was what made the price striking. The other three statements are all accurate.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1eft7xtxthso7" }, { "id": "ot-backers-1", "shape": "mcq", "tags": [ "musk", "openai" ], "prompt": { "modality": "text", "value": "Mallaby treats the October 2012 clash between Musk and Page over advanced AI as planting the seed of which later organisation?" }, "options": [ { "modality": "text", "value": "DeepMind's Ethics and Safety Review Board" }, { "modality": "text", "value": "xAI" }, { "modality": "text", "value": "OpenAI" }, { "modality": "text", "value": "Anthropic" } ], "correctIndex": 2, "explanation": "Mallaby frames that night as the moment the informal Google–Musk alignment cracked, seeding what became OpenAI. The ethics board came out of the Google deal instead, and Anthropic and xAI came years later.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "oivelebtiey8" }, { "id": "ot-backers-2", "shape": "mcq", "tags": [ "musk", "dinner" ], "prompt": { "modality": "text", "value": "When Musk clashed with Page over advanced AI in October 2012, what phrase did he use for a sufficiently advanced AI?" }, "options": [ { "modality": "text", "value": "\"A digital god\"" }, { "modality": "text", "value": "\"A summoned demon\"" }, { "modality": "text", "value": "\"An electric brain\"" }, { "modality": "text", "value": "\"The last invention\"" } ], "correctIndex": 0, "explanation": "Musk warned that advanced AI could become \"a digital god\" and that humanity had no plan for what follows. The other phrases are plausible AI-doom rhetoric but not what the guide records.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "tax3fo1ummrv2" }, { "id": "ot-backers-3", "shape": "mcq", "tags": [ "musk", "larry-page" ], "prompt": { "modality": "text", "value": "Where did the Musk–Page argument over advanced AI take place on 8 October 2012?" }, "options": [ { "modality": "text", "value": "At DeepMind's London office" }, { "modality": "text", "value": "At a tech conference keynote in San Francisco" }, { "modality": "text", "value": "At a dinner at Larry Page's home in Palo Alto" }, { "modality": "text", "value": "At a Google board meeting in Mountain View" } ], "correctIndex": 2, "explanation": "The exchange happened at a dinner at Larry Page's Palo Alto home on 8 October 2012 — the informal setting is part of why Mallaby treats the night as a turning point.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "5muvri1x2fz6k" }, { "id": "ot-backers-4", "shape": "mcq", "tags": [ "larry-page", "openai" ], "prompt": { "modality": "text", "value": "How did Page respond to Musk's warning that evening?" }, "options": [ { "modality": "text", "value": "Advanced AI was decades away and not worth worrying about", "short": "Advanced AI was far off, not a worry" }, { "modality": "text", "value": "The species that built such an intelligence would be its natural heirs", "short": "Its builders would be its natural heirs" }, { "modality": "text", "value": "Only Google could be trusted to build it safely", "short": "Only Google could build it safely" }, { "modality": "text", "value": "Regulation should be written before any lab went further", "short": "Regulate before any lab goes further" } ], "correctIndex": 1, "explanation": "Page argued that a machine intelligence built by humanity would be its rightful heir — a succession, not a threat. Mallaby frames this clash as the seed of what became OpenAI.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "14hb9sy1etn2v4" }, { "id": "ot-dqn-1", "shape": "mcq", "tags": [ "mnih", "atari" ], "prompt": { "modality": "text", "value": "Who led the small team behind DeepMind's Atari DQN agent?" }, "options": [ { "modality": "text", "value": "Mustafa Suleyman" }, { "modality": "text", "value": "Demis Hassabis" }, { "modality": "text", "value": "Shane Legg" }, { "modality": "text", "value": "Vlad Mnih" } ], "correctIndex": 3, "explanation": "Vlad Mnih, trained in reinforcement learning at Toronto and Alberta, led the Atari strike team. Legg, Hassabis and Suleyman are cofounders of the lab, not the leads on this project.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "x0b9nk19184be" }, { "id": "ot-dqn-2", "shape": "mcq", "tags": [ "q-learning", "definition" ], "prompt": { "modality": "text", "value": "In Q-learning, what does an agent estimate?" }, "options": [ { "modality": "text", "value": "The long-term expected reward of each action from each state" }, { "modality": "text", "value": "The shortest path from the current state to the goal state" }, { "modality": "text", "value": "The probability that any given pixel changes next frame" }, { "modality": "text", "value": "How closely its behaviour matches a human demonstrator's" } ], "correctIndex": 0, "explanation": "Q-learning estimates the long-term expected reward (the \"quality\") of every action from every state, refining those estimates through play. Matching a human demonstrator describes imitation learning instead.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "mtz726tk9t68" }, { "id": "ot-dqn-3", "shape": "mcq", "tags": [ "dqn", "definition" ], "prompt": { "modality": "text", "value": "What two components does a Deep Q-Network combine?" }, "options": [ { "modality": "text", "value": "Q-learning with a deep convolutional network that reads raw pixels", "short": "Q-learning + a deep net on raw pixels" }, { "modality": "text", "value": "Supervised learning on human play logs with a scoring heuristic", "short": "Supervised learning on human play logs" }, { "modality": "text", "value": "Q-learning with a hand-coded rules engine per Atari game", "short": "Q-learning + hand-coded game rules" }, { "modality": "text", "value": "A language model with a search tree over future moves", "short": "A language model + a move search tree" } ], "correctIndex": 0, "explanation": "DQN pairs Q-learning with a deep convolutional network reading raw pixels — and adds experience replay for stability. Game-specific hand-coded rules are exactly what the Atari challenge forbade.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "e9ul1m10l71os" }, { "id": "ot-dqn-4", "shape": "mcq", "tags": [ "atari", "reinforcement-learning" ], "prompt": { "modality": "text", "value": "What was the deliberate constraint on the Atari challenge?" }, "options": [ { "modality": "text", "value": "Learning only by watching recordings of expert human players", "short": "Learn only from expert human replays" }, { "modality": "text", "value": "Access to the games' internal memory state rather than the screen", "short": "Access to game memory, not the screen" }, { "modality": "text", "value": "A separate network tuned for each of the Atari 2600 titles", "short": "A separate net tuned per Atari title" }, { "modality": "text", "value": "One network, no game-specific code, learning from raw pixels and the score alone", "short": "One net, no game code, pixels + score" } ], "correctIndex": 3, "explanation": "The constraint — one network, raw pixels, score as the only feedback, no game-specific code — forced a general solution rather than a tailored one, which was the whole point.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1yo4cjpksctwn" }, { "id": "ot-dmdef-1", "shape": "mcq", "tags": [ "deepmind", "definition" ], "prompt": { "modality": "text", "value": "Which description best defines DeepMind?" }, "options": [ { "modality": "text", "value": "A Toronto neural-network research group acquired by Microsoft in 2014", "short": "Toronto group, Microsoft-bought '14" }, { "modality": "text", "value": "The London AI lab Demis Hassabis cofounded in 2010 and Google acquired in 2014", "short": "London lab, Hassabis '10, Google '14" }, { "modality": "text", "value": "Google's internal deep-learning team, formed in Mountain View in 2011", "short": "Google in-house team, Mtn View '11" }, { "modality": "text", "value": "A Silicon Valley AI lab founded by Elon Musk as a nonprofit in 2015", "short": "Silicon Valley, Musk nonprofit, '15" } ], "correctIndex": 1, "explanation": "DeepMind is the London lab Hassabis cofounded in 2010 and Google bought in 2014. The Musk nonprofit describes OpenAI — a different organisation that this lesson traces back to the same 2012 dinner.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "118gzf41f30fli" }, { "id": "ot-dmdef-2", "shape": "mcq", "tags": [ "deepmind", "london" ], "prompt": { "modality": "text", "value": "Which city was DeepMind founded and based in?" }, "options": [ { "modality": "text", "value": "London" }, { "modality": "text", "value": "Mountain View" }, { "modality": "text", "value": "Toronto" }, { "modality": "text", "value": "Palo Alto" } ], "correctIndex": 0, "explanation": "DeepMind is a London company — the acquisition was the largest AI deal in European history. Palo Alto hosted the Musk–Page dinner, and Toronto is where Vlad Mnih trained.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1de9una5da8rk" }, { "id": "ot-dmdef-3", "shape": "mcq", "tags": [ "deepmind", "mission" ], "prompt": { "modality": "text", "value": "How does DeepMind state its mission?" }, "options": [ { "modality": "text", "value": "Organise the world's information and make it universally accessible", "short": "Organise the world's information" }, { "modality": "text", "value": "Build safe machines that never exceed human capability", "short": "Build machines that never exceed humans" }, { "modality": "text", "value": "Solve intelligence, and then use it to solve everything else", "short": "Solve intelligence, then everything else" }, { "modality": "text", "value": "Ensure artificial general intelligence benefits all of humanity", "short": "Ensure AGI benefits all humanity" } ], "correctIndex": 2, "explanation": "DeepMind's stated mission is to solve intelligence and then use it to solve everything else. The second option is OpenAI's mission and the third is Google's — both tempting neighbours.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1cwzafy17ywbe0" }, { "id": "ot-dmdef-4", "shape": "mcq", "tags": [ "deepmind", "definition" ], "prompt": { "modality": "text", "value": "A profile describes \"the London lab behind AlphaGo and AlphaFold, now owned by Google.\" Which organisation is that?" }, "options": [ { "modality": "text", "value": "DeepMind" }, { "modality": "text", "value": "Meta AI (FAIR)" }, { "modality": "text", "value": "OpenAI" }, { "modality": "text", "value": "Google Brain" } ], "correctIndex": 0, "explanation": "AlphaGo and AlphaFold are DeepMind milestones, and the London base plus Google ownership pins it down. Google Brain was a separate, US-based Google team; OpenAI and Meta AI are neither London-based nor Google-owned.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "rgzsx01iuiqq" }, { "id": "ot-dmdef-5", "shape": "mcq", "tags": [ "deepmind", "timeline" ], "prompt": { "modality": "text", "value": "Which pairing of years correctly matches DeepMind's founding and its acquisition by Google?" }, "options": [ { "modality": "text", "value": "Founded 2012, acquired 2014" }, { "modality": "text", "value": "Founded 2011, acquired 2012" }, { "modality": "text", "value": "Founded 2010, acquired 2014" }, { "modality": "text", "value": "Founded 2010, acquired 2011" } ], "correctIndex": 2, "explanation": "Founded in 2010 and acquired by Google in January 2014. 2011 is when the London company filed and 2012 is when Page and Brin started watching — neither is a founding or acquisition year.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1jdqrce5m7kuw" }, { "id": "ot-rl-define-1", "shape": "mcq", "tags": [ "reinforcement-learning", "definition" ], "prompt": { "modality": "text", "value": "Which description best defines reinforcement learning as DeepMind applied it to Atari?" }, "options": [ { "modality": "text", "value": "A network is shown labelled examples and corrected toward the right label", "short": "Corrected toward the right labels" }, { "modality": "text", "value": "Groups are found in unlabelled data by measuring distance between points", "short": "Group unlabelled data by distance" }, { "modality": "text", "value": "An agent learns by trial and error, receiving rewards for good actions and penalties for bad ones", "short": "Trial and error, via reward and penalty" }, { "modality": "text", "value": "A network compresses inputs and reconstructs them to find hidden structure", "short": "Compress inputs, reconstruct structure" } ], "correctIndex": 2, "explanation": "Reinforcement learning is trial-and-error training driven by rewards and penalties — that is how DQN learned Atari from raw pixels. Learning from labelled examples describes supervised learning, not RL.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1q9rhxthxeqz7" }, { "id": "ot-rl-define-2", "shape": "mcq", "tags": [ "dqn", "nature", "dates" ], "prompt": { "modality": "text", "value": "In what year was the Deep Q-Network (DQN) published in Nature?" }, "options": [ { "modality": "text", "value": "2016" }, { "modality": "text", "value": "2012" }, { "modality": "text", "value": "2013" }, { "modality": "text", "value": "2015" } ], "correctIndex": 3, "explanation": "The DQN paper appeared in Nature in February 2015. 2013 is tempting — DeepMind showed early Atari results at NeurIPS that December — but the Nature publication came in 2015.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1vj44nn1xleh3h" }, { "id": "ot-rl-define-3", "shape": "mcq", "tags": [ "dqn", "atari", "numbers" ], "prompt": { "modality": "text", "value": "How many Atari 2600 games was the single DQN architecture retrained on, and on how many did it reach human-level or better?" }, "options": [ { "modality": "text", "value": "29 games, human-level or better on 49" }, { "modality": "text", "value": "49 games, human-level or better on 49" }, { "modality": "text", "value": "37 games, human-level or better on 10" }, { "modality": "text", "value": "49 games, human-level or better on 29" } ], "correctIndex": 3, "explanation": "One architecture was retrained from scratch on 49 games and hit human-level or better on 29 of them. The 29/49 figures are often flipped — the smaller number is the successes.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "xd3sjb1ng4xth" }, { "id": "ot-rl-define-4", "shape": "mcq", "tags": [ "dqn", "generality" ], "prompt": { "modality": "text", "value": "Which of the following was NOT part of what made DQN's Atari result a milestone?" }, "options": [ { "modality": "text", "value": "The same code and hyperparameters used across every game" }, { "modality": "text", "value": "Hand-engineered, game-specific features tuned for each title" }, { "modality": "text", "value": "Raw screen pixels and a score signal as the only input" }, { "modality": "text", "value": "One general algorithm mastering many distinct environments" } ], "correctIndex": 1, "explanation": "The whole point was the absence of hand-engineered per-game features — earlier game-playing AI was hand-tuned per title. DQN used only raw pixels and score, with identical code and hyperparameters throughout.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "a11dmi3huzso" }, { "id": "ot-rl-define-5", "shape": "mcq", "tags": [ "deepmind", "neurips", "dates" ], "prompt": { "modality": "text", "value": "Where and when did DeepMind first demonstrate early Atari results, before the Google acquisition closed?" }, "options": [ { "modality": "text", "value": "NeurIPS, December 2013" }, { "modality": "text", "value": "NeurIPS, December 2015" }, { "modality": "text", "value": "ImageNet competition, 2012" }, { "modality": "text", "value": "The Nature cover, January 2016" } ], "correctIndex": 0, "explanation": "DeepMind showed early Atari results at NeurIPS in December 2013, months ahead of the acquisition. 2012 was AlexNet's ImageNet win, and January 2016 was the AlphaGo Nature cover.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "r5y98p17k87kj" }, { "id": "ot-alphago-define-1", "shape": "mcq", "tags": [ "alphago", "definition" ], "prompt": { "modality": "text", "value": "AlphaGo is best defined as which of the following?" }, "options": [ { "modality": "text", "value": "The DeepMind program that defeated world champion Lee Sedol at Go in 2016", "short": "Beat Lee Sedol at Go in 2016" }, { "modality": "text", "value": "The OpenAI system announced as a counterweight to Google DeepMind", "short": "OpenAI, counterweight to DeepMind" }, { "modality": "text", "value": "The DeepMind network that learned 49 Atari games from raw pixels", "short": "Learned 49 Atari games from pixels" }, { "modality": "text", "value": "The Toronto network that won the ImageNet competition in 2012", "short": "Won ImageNet 2012, from Toronto" } ], "correctIndex": 0, "explanation": "AlphaGo is DeepMind's Go program that beat Lee Sedol in 2016. The Atari-from-pixels system was DQN, a separate DeepMind project.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1jqeyli65jt4g" }, { "id": "ot-alphago-define-2", "shape": "mcq", "tags": [ "alphago", "architecture" ], "prompt": { "modality": "text", "value": "Which two techniques did AlphaGo combine?" }, "options": [ { "modality": "text", "value": "Deep neural networks and Monte Carlo tree search" }, { "modality": "text", "value": "Monte Carlo tree search and brute-force exhaustive search" }, { "modality": "text", "value": "Deep neural networks and linear regression on board features" }, { "modality": "text", "value": "Deep neural networks and hand-coded opening books" } ], "correctIndex": 0, "explanation": "AlphaGo paired deep neural networks with Monte Carlo tree search. Brute-force exhaustive search was impossible — Go's search space had defeated every serious AI approach for decades.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "lpjmud1ft1f7f" }, { "id": "ot-alphago-define-3", "shape": "mcq", "tags": [ "alphago", "lee-sedol", "seoul" ], "prompt": { "modality": "text", "value": "In which city and year did AlphaGo defeat Lee Sedol, and by what score?" }, "options": [ { "modality": "text", "value": "Seoul, March 2015, 5–0" }, { "modality": "text", "value": "London, October 2015, 5–0" }, { "modality": "text", "value": "Seoul, January 2016, 3–2" }, { "modality": "text", "value": "Seoul, March 2016, 4–1" } ], "correctIndex": 3, "explanation": "AlphaGo beat Lee Sedol 4–1 in Seoul in March 2016. The 5–0 sweep was the earlier closed-door match against Fan Hui in October 2015.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "15j9cpbj0b53p" }, { "id": "ot-alphago-define-4", "shape": "mcq", "tags": [ "alphago", "definition" ], "prompt": { "modality": "text", "value": "Which statement about AlphaGo is NOT accurate?" }, "options": [ { "modality": "text", "value": "It combined deep neural networks with Monte Carlo tree search", "short": "Combined neural nets with tree search" }, { "modality": "text", "value": "It learned Go end to end from raw screen pixels alone", "short": "Learned from raw screen pixels alone" }, { "modality": "text", "value": "It was built by DeepMind and defeated Lee Sedol in 2016", "short": "DeepMind built it; beat Lee Sedol, 2016" }, { "modality": "text", "value": "Its architecture was revealed in a January 2016 Nature cover paper", "short": "Revealed in a 2016 Nature cover paper" } ], "correctIndex": 1, "explanation": "Learning from raw pixels with no search describes DQN on Atari, not AlphaGo — AlphaGo's networks worked together with Monte Carlo tree search, and the January 2016 Nature cover revealed that architecture.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1xcffga1jksaco" }, { "id": "ot-go-people-1", "shape": "mcq", "tags": [ "aja-huang", "alphago" ], "prompt": { "modality": "text", "value": "Which engineer built AlphaGo's first prototype, contributing the critical tree-search component?" }, "options": [ { "modality": "text", "value": "Fan Hui" }, { "modality": "text", "value": "Aja Huang" }, { "modality": "text", "value": "Ilya Sutskever" }, { "modality": "text", "value": "Pieter Abbeel" } ], "correctIndex": 1, "explanation": "Aja Huang built AlphaGo's first prototype; his tree-search work made the system competitive at professional level. Fan Hui joined DeepMind later, as an analyst rather than an engineer.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1dqchrwfw6bi" }, { "id": "ot-go-people-2", "shape": "mcq", "tags": [ "sergey-brin", "alphago" ], "prompt": { "modality": "text", "value": "Who asked Hassabis on a May 2014 hallway walk what hard problem DeepMind would tackle next, then funded the answer despite being sceptical?" }, "options": [ { "modality": "text", "value": "Sam Altman" }, { "modality": "text", "value": "Larry Page" }, { "modality": "text", "value": "Sergey Brin" }, { "modality": "text", "value": "Elon Musk" } ], "correctIndex": 2, "explanation": "Sergey Brin posed the question, heard \"Go,\" doubted it was winnable, and provided the resources anyway. Musk and Altman were on the other side of this story — they founded OpenAI as a counterweight.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1y9z8jligojcv" }, { "id": "ot-go-people-3", "shape": "mcq", "tags": [ "fan-hui", "go" ], "prompt": { "modality": "text", "value": "Fan Hui held which title when he played AlphaGo in October 2015?" }, "options": [ { "modality": "text", "value": "Korean national Go champion" }, { "modality": "text", "value": "European Go champion" }, { "modality": "text", "value": "World Go champion" }, { "modality": "text", "value": "Chinese Go champion" } ], "correctIndex": 1, "explanation": "Fan Hui was the European Go champion and lost 5–0 in the closed-door match. Lee Sedol, not Fan Hui, was the world-class player widely called the greatest of his generation.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "11kzwons9ty55" }, { "id": "ot-go-people-4", "shape": "mcq", "tags": [ "lee-sedol", "seoul" ], "prompt": { "modality": "text", "value": "Lee Sedol took one game off AlphaGo in Seoul. Which game was it, and what explains his win?" }, "options": [ { "modality": "text", "value": "Game one — AlphaGo's tree search ran out of thinking time", "short": "Game 1 — its search ran out of time" }, { "modality": "text", "value": "Game two — AlphaGo's shoulder-hit on the fifth line backfired", "short": "Game 2 — its 5th-line shoulder-hit failed" }, { "modality": "text", "value": "Game five — AlphaGo misread the final scoring", "short": "Game 5 — it misread the scoring" }, { "modality": "text", "value": "Game four — he found a move AlphaGo had not adequately evaluated", "short": "Game 4 — a move AlphaGo underevaluated" } ], "correctIndex": 3, "explanation": "Lee Sedol won game four with a move AlphaGo had not foreseen — his lone victory in a 4–1 defeat. Game two belonged to AlphaGo; that was the game containing Move 37, which worked exactly as intended.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "19w5b8q1f4wubo" }, { "id": "ot-go-people-5", "shape": "mcq", "tags": [ "alphago", "people" ], "prompt": { "modality": "text", "value": "Which pairing of person and role is INCORRECT?" }, "options": [ { "modality": "text", "value": "Aja Huang — engineer whose tree-search work proved critical", "short": "Aja Huang — engineer, key tree-search work" }, { "modality": "text", "value": "Lee Sedol — European champion who lost 5–0 in a closed-door match", "short": "Lee Sedol — European champ, lost 5–0" }, { "modality": "text", "value": "Fan Hui — European champion who later joined DeepMind as an analyst", "short": "Fan Hui — European champ, joined DeepMind" }, { "modality": "text", "value": "Sergey Brin — Google cofounder who funded the Go project despite scepticism", "short": "Sergey Brin — Google cofounder, funder" } ], "correctIndex": 1, "explanation": "The 5–0 closed-door loss was Fan Hui's, in October 2015. Lee Sedol was the Korean champion who lost 4–1 in Seoul in March 2016 — a public match, not a closed-door one.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "e4h4ehnaq9kn" }, { "id": "ot-move37-1", "shape": "mcq", "tags": [ "move-37", "alphago" ], "prompt": { "modality": "text", "value": "In which game of the Seoul match, and at what board location, did AlphaGo play the move that became known as Move 37?" }, "options": [ { "modality": "text", "value": "Game four, a shoulder-hit on the fifth line" }, { "modality": "text", "value": "Game two, a shoulder-hit on the fifth line" }, { "modality": "text", "value": "Game two, a corner invasion on the third line" }, { "modality": "text", "value": "Game one, a shoulder-hit on the fourth line" } ], "correctIndex": 1, "explanation": "Move 37 was a shoulder-hit on the fifth line in game two. Game four was the one Lee Sedol won, with a move AlphaGo had not adequately evaluated.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "ou8xz1y7626z" }, { "id": "ot-move37-2", "shape": "mcq", "tags": [ "move-37", "commentary" ], "prompt": { "modality": "text", "value": "How did human grandmasters and commentators initially judge Move 37?" }, "options": [ { "modality": "text", "value": "As a standard professional opening" }, { "modality": "text", "value": "As a mistake" }, { "modality": "text", "value": "As an obvious winning move" }, { "modality": "text", "value": "As a deliberate stalling tactic" } ], "correctIndex": 1, "explanation": "Commentators were stopped cold and rated it a mistake. AlphaGo's reasoning turned out to be correct — the move won territory in a way human intuition would not reach.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "ske8cr1q9cx0l" }, { "id": "ot-move37-3", "shape": "mcq", "tags": [ "move-37", "lee-sedol" ], "prompt": { "modality": "text", "value": "How did Lee Sedol respond immediately after Move 37 was played?" }, "options": [ { "modality": "text", "value": "He resigned the game on the spot", "short": "Resigned the game on the spot" }, { "modality": "text", "value": "He left the board for about twelve minutes and returned visibly shaken", "short": "Left the board ~12 min, shaken" }, { "modality": "text", "value": "He requested a review of AlphaGo's legality", "short": "Asked for a legality review" }, { "modality": "text", "value": "He replied instantly with a move that won the game", "short": "Replied instantly and won" } ], "correctIndex": 1, "explanation": "Lee Sedol spent roughly twelve minutes away from the board and came back visibly shaken. He did win game four later, but game two was AlphaGo's.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1wylkn11jy0f93" }, { "id": "ot-move37-4", "shape": "mcq", "tags": [ "move-37", "ai-creativity" ], "prompt": { "modality": "text", "value": "What has \"Move 37\" come to stand for?" }, "options": [ { "modality": "text", "value": "That AI systems fail in ways humans can easily predict", "short": "AI fails in ways humans can predict" }, { "modality": "text", "value": "That trained networks can discover strategy outside the corpus of human play", "short": "AI can find strategy beyond human play" }, { "modality": "text", "value": "That tree search alone can outperform human intuition", "short": "Tree search alone beats human intuition" }, { "modality": "text", "value": "That neural networks reliably reproduce expert human judgement", "short": "AI just reproduces expert human judgement" } ], "correctIndex": 1, "explanation": "Move 37 is shorthand for networks finding strategic concepts beyond human play. Reproducing expert judgement is the opposite of the point — no human grandmaster would have chosen it.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "w743ra1v5q74g" }, { "id": "ot-openai-founding-1", "shape": "mcq", "tags": [ "openai", "sam-altman" ], "prompt": { "modality": "text", "value": "Which OpenAI cofounder became the organisation's CEO?" }, "options": [ { "modality": "text", "value": "Ilya Sutskever" }, { "modality": "text", "value": "Sam Altman" }, { "modality": "text", "value": "Elon Musk" }, { "modality": "text", "value": "Pieter Abbeel" } ], "correctIndex": 1, "explanation": "Sam Altman became CEO. Musk cofounded and announced OpenAI alongside him, and Sutskever took the chief scientist role rather than the top job.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1cizvcdahiysb" }, { "id": "ot-openai-founding-2", "shape": "mcq", "tags": [ "openai", "ilya-sutskever" ], "prompt": { "modality": "text", "value": "Who became OpenAI's first chief scientist, shaping its early research agenda?" }, "options": [ { "modality": "text", "value": "Sam Altman" }, { "modality": "text", "value": "Ilya Sutskever" }, { "modality": "text", "value": "Pieter Abbeel" }, { "modality": "text", "value": "Geoffrey Hinton" } ], "correctIndex": 1, "explanation": "Ilya Sutskever became OpenAI's first chief scientist. Hinton was his doctoral adviser in Toronto but did not take the role; Abbeel was among the hires who followed later.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1wkcw6x1t7lm9n" }, { "id": "ot-openai-founding-3", "shape": "mcq", "tags": [ "openai", "founding", "funding" ], "prompt": { "modality": "text", "value": "When was OpenAI announced, and how much funding was pledged?" }, "options": [ { "modality": "text", "value": "December 2015, $100 million in commitments" }, { "modality": "text", "value": "May 2014, $10 billion in commitments" }, { "modality": "text", "value": "March 2016, $1 billion in commitments" }, { "modality": "text", "value": "December 2015, $1 billion in commitments" } ], "correctIndex": 3, "explanation": "Musk and Altman announced OpenAI in December 2015 with $1 billion in funding commitments — timing that coincided almost exactly with AlphaGo's closed-door win over Fan Hui.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1fchubr1y32109" }, { "id": "ot-openai-founding-4", "shape": "mcq", "tags": [ "openai", "deepmind", "motivation" ], "prompt": { "modality": "text", "value": "What was the stated rationale for founding OpenAI as a non-profit?" }, "options": [ { "modality": "text", "value": "To act as a counterweight to Google's ownership of DeepMind" }, { "modality": "text", "value": "To commercialise DeepMind's Atari research more quickly" }, { "modality": "text", "value": "To build a Go program that could surpass AlphaGo" }, { "modality": "text", "value": "To fund academic AI departments left behind by industry" } ], "correctIndex": 0, "explanation": "OpenAI was framed explicitly as a counterweight to Google's ownership of DeepMind — Musk saw frontier capability concentrated in one commercial giant as dangerous.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "3y3e9i44f0vg" }, { "id": "ot-openai-founding-5", "shape": "mcq", "tags": [ "openai", "founders", "roles" ], "prompt": { "modality": "text", "value": "Which of the following was NOT part of OpenAI's founding?" }, "options": [ { "modality": "text", "value": "A December 2015 announcement by Elon Musk and Sam Altman", "short": "Dec 2015 Musk & Altman reveal" }, { "modality": "text", "value": "A non-profit structure with $1 billion in pledged commitments", "short": "Non-profit, $1B pledged funding" }, { "modality": "text", "value": "An explicit framing as a counterweight to Google's DeepMind", "short": "A counterweight to DeepMind" }, { "modality": "text", "value": "Ilya Sutskever taking the CEO role while Sam Altman led research", "short": "Sutskever CEO, Altman research" } ], "correctIndex": 3, "explanation": "The roles are reversed: Sam Altman was CEO and Ilya Sutskever was chief scientist. The other three describe the founding accurately.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1l9yneygkv33k" }, { "id": "ot-openai-split-1", "shape": "mcq", "tags": [ "musk", "2017-email", "openai" ], "prompt": { "modality": "text", "value": "Who sent the 20 September 2017 memo to Musk and Altman warning that one person could end up with \"absolute control\" of AI?" }, "options": [ { "modality": "text", "value": "Dario Amodei and Paul Christiano" }, { "modality": "text", "value": "Sundar Pichai and Demis Hassabis" }, { "modality": "text", "value": "Shane Legg and Mustafa Suleyman" }, { "modality": "text", "value": "Greg Brockman and Ilya Sutskever" } ], "correctIndex": 3, "explanation": "Brockman and Sutskever — both OpenAI founders — wrote the memo. Amodei and Christiano were also at OpenAI, but their break came later, in the 2021 walkout that founded Anthropic.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1rx0fqz1ywf4np" }, { "id": "ot-openai-split-2", "shape": "mcq", "tags": [ "2017-email", "hassabis" ], "prompt": { "modality": "text", "value": "The September 2017 email named one specific person as the target of its \"absolute control\" concern. Who was named?" }, "options": [ { "modality": "text", "value": "Sam Altman" }, { "modality": "text", "value": "Satya Nadella" }, { "modality": "text", "value": "Demis Hassabis" }, { "modality": "text", "value": "Larry Page" } ], "correctIndex": 2, "explanation": "Hassabis was the person the memo named — Sutskever wrote of Musk's fear that \"Demis could create an AGI dictatorship.\" Altman ran OpenAI but was a recipient of the warning, not its named target.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1pqvjdv1e2mwed" }, { "id": "ot-openai-split-3", "shape": "mcq", "tags": [ "musk", "openai", "control" ], "prompt": { "modality": "text", "value": "What did Musk push for at OpenAI before his 2018 departure?" }, "options": [ { "modality": "text", "value": "To install Brockman as sole CEO" }, { "modality": "text", "value": "To merge OpenAI into DeepMind" }, { "modality": "text", "value": "To run OpenAI himself" }, { "modality": "text", "value": "To convert OpenAI into a capped-profit entity" } ], "correctIndex": 2, "explanation": "Musk pushed to run OpenAI himself and left when he failed to gain operational control. The capped-profit conversion came in 2019, after he had already resigned.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "23f75z1fd7q1p" }, { "id": "ot-openai-split-4", "shape": "mcq", "tags": [ "2017-email", "concentration" ], "prompt": { "modality": "text", "value": "Why does the guide call the September 2017 email significant?" }, "options": [ { "modality": "text", "value": "It was the first written warning that AGI could arrive before 2030", "short": "First written warning: AGI before 2030" }, { "modality": "text", "value": "It showed founders of an anti-monopoly lab feared the very concentration they were enacting", "short": "Founders feared their own concentration" }, { "modality": "text", "value": "It proved Musk had already decided to withdraw his funding", "short": "Musk decided to withdraw funding" }, { "modality": "text", "value": "It triggered Google Brain's merger into DeepMind", "short": "Triggered Google Brain–DeepMind merger" } ], "correctIndex": 1, "explanation": "The email's significance is the irony: OpenAI was founded to prevent an AI monopoly, yet its own founders feared the logic of concentration inside their lab. Musk's withdrawal came later, in early 2018.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "6hunj6v2g6v0" }, { "id": "ot-capped-profit-1", "shape": "mcq", "tags": [ "capped-profit", "100x" ], "prompt": { "modality": "text", "value": "In OpenAI's 2019 restructuring, early investor returns were capped at what multiple of their stake?" }, "options": [ { "modality": "text", "value": "100 times" }, { "modality": "text", "value": "10 times" }, { "modality": "text", "value": "50 times" }, { "modality": "text", "value": "1,000 times" } ], "correctIndex": 0, "explanation": "The cap was set at 100× an investor's stake — profits would be real but bounded. The figure later became a flashpoint the board cited against Altman.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "hs7dfg1bv18k2" }, { "id": "ot-capped-profit-2", "shape": "mcq", "tags": [ "capped-profit", "2019" ], "prompt": { "modality": "text", "value": "In what year did OpenAI restructure into a capped-profit company?" }, "options": [ { "modality": "text", "value": "2017" }, { "modality": "text", "value": "2023" }, { "modality": "text", "value": "2021" }, { "modality": "text", "value": "2019" } ], "correctIndex": 3, "explanation": "The restructuring happened in 2019. 2017 was the Brockman-Sutskever email, 2021 the Anthropic split, and 2023 the Altman firing.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "fc4jce17pkhso" }, { "id": "ot-capped-profit-3", "shape": "mcq", "tags": [ "capped-profit", "governance" ], "prompt": { "modality": "text", "value": "What corporate structure did OpenAI adopt in 2019?" }, "options": [ { "modality": "text", "value": "A public benefit corporation wholly owned by Microsoft" }, { "modality": "text", "value": "A for-profit parent with a nonprofit research arm" }, { "modality": "text", "value": "A nonprofit parent controlling a for-profit subsidiary" }, { "modality": "text", "value": "A joint venture split evenly with Microsoft" } ], "correctIndex": 2, "explanation": "The nonprofit parent retained formal control over a for-profit subsidiary — the very structure that left a small nonprofit board able to fire Altman in 2023.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1wq70vk5mzrtq" }, { "id": "ot-capped-profit-4", "shape": "mcq", "tags": [ "microsoft", "funding", "capped-profit" ], "prompt": { "modality": "text", "value": "What did Altman secure using the new 2019 structure?" }, "options": [ { "modality": "text", "value": "A $1 billion commitment from Google" }, { "modality": "text", "value": "A $100 million commitment from Musk" }, { "modality": "text", "value": "A $1 billion commitment from Microsoft" }, { "modality": "text", "value": "A $10 billion commitment from Microsoft" } ], "correctIndex": 2, "explanation": "The capped-profit structure let Altman raise $1 billion from Microsoft — the partnership that would make Microsoft the dominant AI-infrastructure player.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "u5oa3r1pk2vq9" }, { "id": "ot-capped-profit-5", "shape": "mcq", "tags": [ "capped-profit", "microsoft" ], "prompt": { "modality": "text", "value": "Which statement about the 2019 capped-profit pivot is NOT accurate?" }, "options": [ { "modality": "text", "value": "Profits would be real but bounded", "short": "Profits would be real but bounded" }, { "modality": "text", "value": "The board later cited the cap as a commitment Altman had eroded", "short": "Board: cap a commitment Altman eroded" }, { "modality": "text", "value": "The cap was presented as a way to end Microsoft's influence over OpenAI", "short": "Framed as ending Microsoft's sway" }, { "modality": "text", "value": "The move was presented as a safety compromise", "short": "Framed as a safety compromise" } ], "correctIndex": 2, "explanation": "The opposite is true: the structure is what enabled Microsoft's $1 billion commitment and its growing influence. The other three all describe the pivot accurately.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1ydz4x516ksucv" }, { "id": "ot-gdm-merger-1", "shape": "mcq", "tags": [ "google-deepmind", "merger", "2023" ], "prompt": { "modality": "text", "value": "Which two units merged in April 2023 to form Google DeepMind?" }, "options": [ { "modality": "text", "value": "Google Brain and DeepMind" }, { "modality": "text", "value": "DeepMind and Microsoft AI" }, { "modality": "text", "value": "Google Brain and Google X" }, { "modality": "text", "value": "Google Research and DeepMind" } ], "correctIndex": 0, "explanation": "Google Brain merged with DeepMind into a single unit. Google Brain was the separate faction whose decade-long rivalry with DeepMind the merger ended.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "utj4dzzczeod" }, { "id": "ot-gdm-merger-2", "shape": "mcq", "tags": [ "hassabis", "google-deepmind" ], "prompt": { "modality": "text", "value": "Who became CEO of the merged Google DeepMind?" }, "options": [ { "modality": "text", "value": "Shane Legg" }, { "modality": "text", "value": "Sundar Pichai" }, { "modality": "text", "value": "Demis Hassabis" }, { "modality": "text", "value": "Mustafa Suleyman" } ], "correctIndex": 2, "explanation": "Hassabis led the merged unit, handing him direct control over Google's entire response to OpenAI. Pichai ordered the merger but ran Alphabet; Legg stayed DeepMind's chief AGI scientist, not its CEO.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "wc8poj1selddd" }, { "id": "ot-gdm-merger-3", "shape": "mcq", "tags": [ "chatgpt", "merger", "pichai" ], "prompt": { "modality": "text", "value": "What prompted Sundar Pichai's April 2023 consolidation of Google's AI teams?" }, "options": [ { "modality": "text", "value": "ChatGPT's launch in late 2022 shocked Alphabet" }, { "modality": "text", "value": "Microsoft's $1 billion commitment to OpenAI" }, { "modality": "text", "value": "Musk's resignation from the OpenAI board" }, { "modality": "text", "value": "Suleyman's 2019 departure to Google" } ], "correctIndex": 0, "explanation": "ChatGPT's late-2022 launch sent shockwaves through Alphabet, and the merger was Pichai's April 2023 response. The Microsoft commitment dated to 2019 and did not trigger the merger.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "c9vn2qqcbqws" }, { "id": "ot-gdm-merger-4", "shape": "mcq", "tags": [ "transformer", "google-brain" ], "prompt": { "modality": "text", "value": "Which architecture, invented in 2017, came from the team that merged into DeepMind?" }, "options": [ { "modality": "text", "value": "Deep reinforcement learning" }, { "modality": "text", "value": "The transformer" }, { "modality": "text", "value": "The convolutional neural network" }, { "modality": "text", "value": "Constitutional AI" } ], "correctIndex": 1, "explanation": "Google Brain invented the transformer in 2017. Reinforcement learning was DeepMind's own tradition — the merger finally put both sets of researchers in the same room.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "vwft0exln4hs" }, { "id": "ot-gdm-merger-5", "shape": "mcq", "tags": [ "merger", "rivalry" ], "prompt": { "modality": "text", "value": "According to the guide, what did the April 2023 merger end?" }, "options": [ { "modality": "text", "value": "DeepMind's independence from Alphabet ownership" }, { "modality": "text", "value": "Alphabet's funding of external AI labs" }, { "modality": "text", "value": "Google's use of the transformer architecture" }, { "modality": "text", "value": "A decade of internal rivalry between two Google AI factions" } ], "correctIndex": 3, "explanation": "The merger closed a decade-long rivalry between Google Brain and DeepMind. DeepMind was already Alphabet-owned, and the transformer remained central rather than abandoned.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1uvi6z51ckaq63" }, { "id": "ot-three-giants-1", "shape": "mcq", "tags": [ "legg", "deepmind", "2024" ], "prompt": { "modality": "text", "value": "By 2024, what role did Shane Legg hold?" }, "options": [ { "modality": "text", "value": "DeepMind's chief AGI scientist" }, { "modality": "text", "value": "CEO of Google DeepMind" }, { "modality": "text", "value": "Head of the UK AI Safety Institute" }, { "modality": "text", "value": "CEO of Microsoft AI" } ], "correctIndex": 0, "explanation": "Legg remained DeepMind's chief AGI scientist while Hassabis ran Google DeepMind and Suleyman ran Microsoft AI — all three co-founders atop major AI operations.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "14823vh1i3mopz" }, { "id": "ot-three-giants-2", "shape": "mcq", "tags": [ "altman", "crisis", "timeline" ], "prompt": { "modality": "text", "value": "How many days passed between Sam Altman's firing and his reinstatement?" }, "options": [ { "modality": "text", "value": "5" }, { "modality": "text", "value": "36" }, { "modality": "text", "value": "2" }, { "modality": "text", "value": "10" } ], "correctIndex": 0, "explanation": "The board folded within 5 days. The 48-hour mark is when 700+ staff signed the resignation letter, and 36 hours is when Nadella made his public offer.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "19ookj0htp5le" }, { "id": "ot-three-giants-3", "shape": "mcq", "tags": [ "nadella", "microsoft", "crisis" ], "prompt": { "modality": "text", "value": "Who publicly offered to create a new Microsoft AI lab for Altman within 36 hours of his dismissal?" }, "options": [ { "modality": "text", "value": "Mustafa Suleyman" }, { "modality": "text", "value": "Satya Nadella" }, { "modality": "text", "value": "Sundar Pichai" }, { "modality": "text", "value": "Greg Brockman" } ], "correctIndex": 1, "explanation": "Nadella made the offer — a calculated use of Microsoft's leverage as OpenAI's cloud provider and largest backer. Suleyman ran Microsoft AI but was not the one who extended the lifeline.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "gbx6gr18aia0x" }, { "id": "ot-three-giants-4", "shape": "mcq", "tags": [ "crisis", "employees", "openai" ], "prompt": { "modality": "text", "value": "Roughly how many of OpenAI's 770 employees signed the letter threatening to resign?" }, "options": [ { "modality": "text", "value": "About 250" }, { "modality": "text", "value": "All 770" }, { "modality": "text", "value": "More than 700" }, { "modality": "text", "value": "About 400" } ], "correctIndex": 2, "explanation": "More than 700 of 770 signed within 48 hours — near-total, but not unanimous, which made the board's position untenable.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1vc3mhvmksj9" }, { "id": "ot-folding-1", "shape": "mcq", "tags": [ "protein-folding", "definition" ], "prompt": { "modality": "text", "value": "Protein folding is the process by which a chain of amino acids does what?" }, "options": [ { "modality": "text", "value": "Twists itself into a specific three-dimensional shape that determines the protein's function", "short": "Twists into a functional 3D shape" }, { "modality": "text", "value": "Splits into shorter peptides that the cell recycles for energy", "short": "Splits into peptides for energy" }, { "modality": "text", "value": "Binds to a crystallography plate so its atoms can be imaged", "short": "Binds to a crystallography plate" }, { "modality": "text", "value": "Copies itself onto a matching strand of messenger RNA", "short": "Copies onto matching messenger RNA" } ], "correctIndex": 0, "explanation": "Protein folding is a chain of amino acids twisting into a specific 3D shape, and that shape determines the protein's function. Crystallography is how the shape is measured experimentally, not what folding is.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1n96i00mymg4y" }, { "id": "ot-folding-2", "shape": "mcq", "tags": [ "anfinsen", "conjecture" ], "prompt": { "modality": "text", "value": "What did Christian Anfinsen conjecture about proteins?" }, "options": [ { "modality": "text", "value": "That extra cellular machinery is required to encode the folded shape", "short": "Extra machinery encodes the shape" }, { "modality": "text", "value": "That a protein's shape can only be determined by crystallography", "short": "Only crystallography reveals shape" }, { "modality": "text", "value": "That most proteins can fold into several equally stable shapes", "short": "Proteins have several stable shapes" }, { "modality": "text", "value": "That a protein's sequence alone determines its folded shape", "short": "Sequence alone sets the folded shape" } ], "correctIndex": 3, "explanation": "Anfinsen's conjecture: the information needed for the folded shape is all in the amino-acid sequence — no additional cellular machinery required. That is the opposite of the second option.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "58izjzbolm3l" }, { "id": "ot-folding-3", "shape": "mcq", "tags": [ "anfinsen", "1972" ], "prompt": { "modality": "text", "value": "Where and when did Anfinsen state his conjecture?" }, "options": [ { "modality": "text", "value": "In his 1972 Nobel Prize lecture" }, { "modality": "text", "value": "In his 2004 memoir on structural biology" }, { "modality": "text", "value": "In a 1994 keynote at the first CASP contest" }, { "modality": "text", "value": "In a 1962 Nobel lecture on the double helix" } ], "correctIndex": 0, "explanation": "Anfinsen stated the conjecture in his 1972 Nobel Prize lecture. CASP began in 1994 under John Moult — a different milestone entirely.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1s2zyqf1k87lxp" }, { "id": "ot-folding-4", "shape": "mcq", "tags": [ "anfinsen", "not-true" ], "prompt": { "modality": "text", "value": "Which statement about Anfinsen's conjecture is NOT true?" }, "options": [ { "modality": "text", "value": "Biologists accepted the principle but could not compute a shape from a sequence", "short": "Biologists accepted but couldn't compute it" }, { "modality": "text", "value": "Biologists rejected the principle and worked to disprove it for fifty years", "short": "Biologists rejected it for fifty years" }, { "modality": "text", "value": "It concerned the information encoded in the amino acids", "short": "It concerned info in amino acids" }, { "modality": "text", "value": "AlphaFold was the algorithmic vindication of the conjecture", "short": "AlphaFold vindicated the conjecture" } ], "correctIndex": 1, "explanation": "Biologists accepted Anfinsen's principle for half a century — they simply could not extract the shape from the sequence in practice. They did not reject it.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1k3ejeh1yrh2dj" }, { "id": "ot-folding-5", "shape": "mcq", "tags": [ "anfinsen", "timeline" ], "prompt": { "modality": "text", "value": "Roughly how long after Anfinsen stated his conjecture did AlphaFold vindicate it algorithmically?" }, "options": [ { "modality": "text", "value": "About twenty-five years" }, { "modality": "text", "value": "About ten years" }, { "modality": "text", "value": "About fifty years" }, { "modality": "text", "value": "About a century" } ], "correctIndex": 2, "explanation": "Anfinsen conjectured in 1972 and AlphaFold delivered around 2020 — roughly fifty years, matching the \"fifty-year grand challenge\" framing used throughout the account.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "12qhad2rs3n30" }, { "id": "ot-afdef-1", "shape": "mcq", "tags": [ "alphafold", "definition" ], "prompt": { "modality": "text", "value": "AlphaFold is the DeepMind system that predicts what, from what input?" }, "options": [ { "modality": "text", "value": "A drug molecule's binding affinity, from a protein's known shape", "short": "Binding affinity from a protein's shape" }, { "modality": "text", "value": "A protein's amino-acid sequence, from a crystallography image", "short": "Sequence from a crystallography image" }, { "modality": "text", "value": "A protein's three-dimensional structure, from its amino-acid sequence", "short": "3D structure from amino-acid sequence" }, { "modality": "text", "value": "A gene's mutation rate, from its evolutionary history", "short": "Mutation rate from evolutionary history" } ], "correctIndex": 2, "explanation": "AlphaFold takes an amino-acid sequence and predicts the 3D structure. The reverse direction — sequence from an image — is not what it does.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1w3k5875hh1vx" }, { "id": "ot-afdef-2", "shape": "mcq", "tags": [ "jumper", "alphafold-2" ], "prompt": { "modality": "text", "value": "Who was the lead scientist on AlphaFold 2?" }, "options": [ { "modality": "text", "value": "Andrew Senior" }, { "modality": "text", "value": "John Moult" }, { "modality": "text", "value": "David Silver" }, { "modality": "text", "value": "John Jumper" } ], "correctIndex": 3, "explanation": "John Jumper led AlphaFold 2. Andrew Senior was the project's first lead and the person who recruited Jumper; Moult ran CASP; Silver led AlphaGo work.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1ks61vtyq387r" }, { "id": "ot-afdef-3", "shape": "mcq", "tags": [ "alphafold", "apply" ], "prompt": { "modality": "text", "value": "A lab has the amino-acid chain of a newly discovered protein but no experimental structure. What would running AlphaFold on it give them?" }, "options": [ { "modality": "text", "value": "A crystallography image of the protein's atoms" }, { "modality": "text", "value": "The gene sequence that coded for the chain" }, { "modality": "text", "value": "A predicted three-dimensional structure for that chain" }, { "modality": "text", "value": "A ranked list of drugs that bind the protein" } ], "correctIndex": 2, "explanation": "AlphaFold's job is sequence in, predicted 3D structure out. It does not produce experimental images, work backwards to genes, or screen drugs — those are downstream or different tools.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "bqjvc9mud0zf" }, { "id": "ot-afdef-4", "shape": "mcq", "tags": [ "jumper", "nobel" ], "prompt": { "modality": "text", "value": "Which prize did John Jumper share with Demis Hassabis for the AlphaFold work?" }, "options": [ { "modality": "text", "value": "The 2022 Breakthrough Prize in Mathematics" }, { "modality": "text", "value": "The 2024 Nobel Prize in Chemistry" }, { "modality": "text", "value": "The 2020 Turing Award" }, { "modality": "text", "value": "The 2024 Nobel Prize in Physiology or Medicine" } ], "correctIndex": 1, "explanation": "Jumper and Hassabis shared the 2024 Nobel Prize in Chemistry for AlphaFold. It was Chemistry, not Physiology or Medicine.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "104lzz9sjkko7" }, { "id": "ot-afdef-5", "shape": "mcq", "tags": [ "jumper", "not-true" ], "prompt": { "modality": "text", "value": "Which of these is NOT an accurate description of AlphaFold or its team?" }, "options": [ { "modality": "text", "value": "AlphaFold predicts 3D structure from an amino-acid sequence", "short": "AlphaFold: sequence to 3D structure" }, { "modality": "text", "value": "Jumper was recruited in October 2017 by Andrew Senior, the project's first lead", "short": "Andrew Senior recruited Jumper, first lead" }, { "modality": "text", "value": "Jumper brought biological intuition the team had lacked", "short": "Jumper brought biological intuition" }, { "modality": "text", "value": "Jumper was recruited in October 2017 by John Moult, who then led AlphaFold 2", "short": "John Moult recruited Jumper, led AlphaFold 2" } ], "correctIndex": 3, "explanation": "Andrew Senior — not John Moult — recruited Jumper in October 2017. Moult was the professor who ran the independent CASP contest.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "offi7s8grqpy" }, { "id": "ot-disto-1", "shape": "mcq", "tags": [ "distogram", "definition" ], "prompt": { "modality": "text", "value": "A \"distogram\" records what information about pairs of amino acids?" }, "options": [ { "modality": "text", "value": "The distance between them on a continuous scale" }, { "modality": "text", "value": "The bond angle between their side chains" }, { "modality": "text", "value": "Their order along the amino-acid chain" }, { "modality": "text", "value": "Whether they touch, as a yes-or-no flag" } ], "correctIndex": 0, "explanation": "A distogram records the continuous distance between amino-acid pairs. The binary touch/no-touch flag is the older contact map that rival labs predicted.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1t1fgcw1m767ce" }, { "id": "ot-disto-2", "shape": "mcq", "tags": [ "contact-map", "rivals" ], "prompt": { "modality": "text", "value": "What did rival labs train their networks to predict about two amino acids?" }, "options": [ { "modality": "text", "value": "A binary label for contact between them" }, { "modality": "text", "value": "The exact 3D coordinates of every atom" }, { "modality": "text", "value": "The evolutionary age of each residue" }, { "modality": "text", "value": "The continuous distance separating them" } ], "correctIndex": 0, "explanation": "Rivals predicted a binary contact map. Predicting continuous distance was AlphaFold's departure; predicting exact 3D coordinates directly came later, with AlphaFold 2's pivot.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1bqe3cucr654c" }, { "id": "ot-disto-3", "shape": "mcq", "tags": [ "distogram", "analogy" ], "prompt": { "modality": "text", "value": "A team member compared the shift from contact maps to distograms to which change?" }, "options": [ { "modality": "text", "value": "Going from sheet music to a live orchestra" }, { "modality": "text", "value": "Going from a telescope to a microscope" }, { "modality": "text", "value": "Going from a paper map to satellite navigation" }, { "modality": "text", "value": "Going from black-and-white to full-colour television" } ], "correctIndex": 3, "explanation": "Researcher Marek Barwinski described it as \"like going from black and white to a full-colour TV\" — the distogram carried far richer geometric information than a binary contact map.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "n20i25xcl4bn" }, { "id": "ot-disto-4", "shape": "mcq", "tags": [ "distogram", "architecture" ], "prompt": { "modality": "text", "value": "Which statement about AlphaFold 1's distogram leap is NOT true?" }, "options": [ { "modality": "text", "value": "It was produced by a convolutional network" }, { "modality": "text", "value": "It gave the model far richer geometric information" }, { "modality": "text", "value": "It was produced by a transformer network built in 2019" }, { "modality": "text", "value": "It was AlphaFold 1's defining architectural choice" } ], "correctIndex": 2, "explanation": "AlphaFold 1's distogram came from a convolutional network. The transformer (the \"tetraformer\") arrived in 2019 with AlphaFold 2 — a later, separate rebuild.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "mbdimi1608obs" }, { "id": "ot-gdtcasp-1", "shape": "mcq", "tags": [ "gdt", "definition" ], "prompt": { "modality": "text", "value": "What does the GDT score measure?" }, "options": [ { "modality": "text", "value": "How many amino acids a predicted protein contains", "short": "How many amino acids a protein has" }, { "modality": "text", "value": "How fast a model produces a structure prediction", "short": "How fast a model predicts structure" }, { "modality": "text", "value": "How many labs entered a given CASP contest", "short": "How many labs entered the CASP contest" }, { "modality": "text", "value": "How closely a prediction's atom positions match the experimentally verified structure", "short": "How well atoms match the true structure" } ], "correctIndex": 3, "explanation": "GDT measures the percentage of a predicted protein's main-chain atoms lying within a tolerated distance of the ground-truth structure — an accuracy score, not a speed or size measure.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1u5uhjv9qc2b9" }, { "id": "ot-gdtcasp-2", "shape": "mcq", "tags": [ "casp", "acronym" ], "prompt": { "modality": "text", "value": "What does the acronym CASP stand for?" }, "options": [ { "modality": "text", "value": "Comparative Assay of Structural Proteins" }, { "modality": "text", "value": "Critical Assessment of Structure Prediction" }, { "modality": "text", "value": "Computational Analysis of Sequence Properties" }, { "modality": "text", "value": "Consortium for Applied Structural Physics" } ], "correctIndex": 1, "explanation": "CASP is the Critical Assessment of Structure Prediction — the contest where protein-folding methods compete on structures known only to the experimentalists.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1q07b8ge7zmnm" }, { "id": "ot-gdtcasp-3", "shape": "mcq", "tags": [ "casp", "cadence" ], "prompt": { "modality": "text", "value": "How often has CASP run, and since when?" }, "options": [ { "modality": "text", "value": "Every two years since 1972" }, { "modality": "text", "value": "Every two years since 1994" }, { "modality": "text", "value": "Every year since 1994" }, { "modality": "text", "value": "Every four years since 2004" } ], "correctIndex": 1, "explanation": "Professor John Moult has run CASP every two years since 1994 — it is biennial. 1972 is the year of Anfinsen's Nobel lecture, a different milestone.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1qvu4zy1ms5c84" }, { "id": "ot-gdtcasp-4", "shape": "mcq", "tags": [ "gdt", "threshold" ], "prompt": { "modality": "text", "value": "Roughly what GDT value marks the threshold at which a prediction is as good as an experimental measurement?" }, "options": [ { "modality": "text", "value": "About 90" }, { "modality": "text", "value": "About 50" }, { "modality": "text", "value": "About 20" }, { "modality": "text", "value": "About 60" } ], "correctIndex": 0, "explanation": "Roughly 90 GDT is the threshold where a prediction is as accurate as experiment. 60 and 20 were mid-pivot AlphaFold 2 scores, not the experimental-quality bar.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1d2plt2fipu3s" }, { "id": "ot-gdtcasp-5", "shape": "mcq", "tags": [ "casp", "not-true" ], "prompt": { "modality": "text", "value": "Which statement about CASP is NOT true?" }, "options": [ { "modality": "text", "value": "Competing labs are given the true shapes in advance so they can calibrate", "short": "Labs get the true shapes in advance" }, { "modality": "text", "value": "Competing labs receive only amino-acid sequences", "short": "Labs receive only amino-acid sequences" }, { "modality": "text", "value": "Predictions are scored by Global Distance Test", "short": "Scored by Global Distance Test" }, { "modality": "text", "value": "The true shapes are known only to the experimentalists", "short": "Only experimentalists know true shapes" } ], "correctIndex": 0, "explanation": "Labs receive only the sequences; the true shapes are withheld and known only to the experimentalists. That blinding is what made CASP an external, objective scoreboard.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "bspvce1ybznpg" }, { "id": "ot-casp14-1", "shape": "mcq", "tags": [ "casp14", "gdt" ], "prompt": { "modality": "text", "value": "What GDT score did AlphaFold 2 achieve at CASP14?" }, "options": [ { "modality": "text", "value": "20" }, { "modality": "text", "value": "92" }, { "modality": "text", "value": "60" }, { "modality": "text", "value": "43" } ], "correctIndex": 1, "explanation": "AlphaFold 2 scored roughly 92 GDT (92.4) at CASP14 in November 2020. 60 and 20 were scores during the direct-folding pivot; 43 was the count of hard cases at CASP13.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1flr2nm1xgtu00" }, { "id": "ot-casp14-2", "shape": "mcq", "tags": [ "casp14", "margin" ], "prompt": { "modality": "text", "value": "How did AlphaFold 2's CASP14 result compare with the contest's previous best?" }, "options": [ { "modality": "text", "value": "About 10 percent higher than the previous best" }, { "modality": "text", "value": "More than 50 percent higher than any previous best" }, { "modality": "text", "value": "Double the previous best, but only in free modelling" }, { "modality": "text", "value": "Roughly tied with the previous best" } ], "correctIndex": 1, "explanation": "AlphaFold 2's 92 GDT was more than 50 percent above any previous best in CASP's history — accurate enough that experimentalists' gold-standard test could not detect a meaningful difference.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "jsxwf51ewt7p3" }, { "id": "ot-casp14-3", "shape": "mcq", "tags": [ "database", "200-million" ], "prompt": { "modality": "text", "value": "Approximately how many proteins did the AlphaFold Structure Database hold by July 2022?" }, "options": [ { "modality": "text", "value": "3 million" }, { "modality": "text", "value": "20,000" }, { "modality": "text", "value": "200 million" }, { "modality": "text", "value": "97" } ], "correctIndex": 2, "explanation": "By July 2022 the database held roughly 200 million entries. 20,000 is the human proteome it covered within months; 3 million is the number of researchers who had used it by 2025.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1lejza1j65040" }, { "id": "ot-casp14-4", "shape": "mcq", "tags": [ "casp14", "not-true" ], "prompt": { "modality": "text", "value": "Which statement about AlphaFold 2's landmark results is NOT true?" }, "options": [ { "modality": "text", "value": "AlphaFold 2 scored about 92 GDT at CASP14", "short": "AlphaFold 2 scored ~92 GDT at CASP14" }, { "modality": "text", "value": "The database held roughly 200 million proteins by July 2022", "short": "Database held ~200M proteins by Jul 2022" }, { "modality": "text", "value": "Moult tallied the winning CASP scores in November 2018", "short": "Moult tallied CASP scores in Nov 2018" }, { "modality": "text", "value": "The direct-folding pivot first sank GDT from about 60 to about 20", "short": "Direct-folding pivot: GDT ~60 to ~20" } ], "correctIndex": 2, "explanation": "Moult tallied CASP14 in November 2020, not 2018. December 2018 was CASP13 in Cancún — the earlier AlphaFold 1 win, a different contest.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "7ilk1mksgovw" }, { "id": "ot-chatgpt-shock-1", "shape": "mcq", "tags": [ "chatgpt", "mallaby" ], "prompt": { "modality": "text", "value": "For Mallaby, what did ChatGPT's launch mark?" }, "options": [ { "modality": "text", "value": "OpenAI's abandonment of its research-preview model" }, { "modality": "text", "value": "The first time deep learning beat a human benchmark" }, { "modality": "text", "value": "The moment the AI race became public" }, { "modality": "text", "value": "The invention of the Transformer architecture" } ], "correctIndex": 2, "explanation": "Mallaby treats the launch as the moment the AI race went public: investors demanded weekly briefings and politicians began scheduling hearings. The Transformer had been invented five years earlier, in 2017.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1tgtiwv1omfpoh" }, { "id": "ot-chatgpt-shock-2", "shape": "mcq", "tags": [ "transformer", "dates" ], "prompt": { "modality": "text", "value": "In what year was the Transformer paper — the architecture ChatGPT ultimately rests on — published?" }, "options": [ { "modality": "text", "value": "2012" }, { "modality": "text", "value": "2017" }, { "modality": "text", "value": "2020" }, { "modality": "text", "value": "2015" } ], "correctIndex": 1, "explanation": "The Transformer paper was published in 2017, five years before ChatGPT's 2022 launch put the architecture in front of the general public.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "jtgrhg1vpjc4q" }, { "id": "ot-chatgpt-shock-3", "shape": "mcq", "tags": [ "chatgpt", "adoption" ], "prompt": { "modality": "text", "value": "ChatGPT reached its first million users in how long after launch?" }, "options": [ { "modality": "text", "value": "Five weeks" }, { "modality": "text", "value": "Five days" }, { "modality": "text", "value": "Five months" }, { "modality": "text", "value": "Five hours" } ], "correctIndex": 1, "explanation": "A million users arrived within five days; 100 million took two months. The five-day figure is the first-million milestone, not the 100-million one.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "126g82ex06fpw" }, { "id": "ot-chatgpt-shock-4", "shape": "mcq", "tags": [ "chatgpt", "openai" ], "prompt": { "modality": "text", "value": "ChatGPT launched as a polished interface on top of which underlying model?" }, "options": [ { "modality": "text", "value": "GPT-2" }, { "modality": "text", "value": "GPT-3.5" }, { "modality": "text", "value": "Gemini" }, { "modality": "text", "value": "GPT-4" } ], "correctIndex": 1, "explanation": "ChatGPT was a research preview wrapping GPT-3.5. GPT-2 was an earlier OpenAI model, GPT-4 arrived after the launch, and Gemini was Google's — none underpinned the November 2022 release.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1huiql91eile5v" }, { "id": "ot-alignment-figures-1", "shape": "mcq", "tags": [ "alignment", "definition" ], "prompt": { "modality": "text", "value": "In the lesson, what is 'alignment'?" }, "options": [ { "modality": "text", "value": "Ensuring an AI reliably pursues human-intended goals as it grows more capable", "short": "AI reliably pursues human-intended goals" }, { "modality": "text", "value": "Tuning chip supply to match a lab's compute demand", "short": "Tuning chip supply to compute demand" }, { "modality": "text", "value": "Coordinating release dates across competing AI labs", "short": "Coordinating release dates across labs" }, { "modality": "text", "value": "Matching a model's training data to its benchmark test set", "short": "Matching training data to the test set" } ], "correctIndex": 0, "explanation": "Alignment is the technical and conceptual problem of an AI reliably pursuing human-intended goals even as capability grows. The other options describe engineering or business coordination, not goal fidelity.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "fcnlk41o8muvu" }, { "id": "ot-alignment-figures-2", "shape": "mcq", "tags": [ "irving", "deepmind" ], "prompt": { "modality": "text", "value": "Which researcher moved from OpenAI to DeepMind in autumn 2019 with an explicit alignment brief, and later became a key figure at the UK AI Safety Institute?" }, "options": [ { "modality": "text", "value": "Paul Christiano" }, { "modality": "text", "value": "Demis Hassabis" }, { "modality": "text", "value": "Geoffrey Irving" }, { "modality": "text", "value": "Dario Amodei" } ], "correctIndex": 2, "explanation": "Geoffrey Irving made that move. Amodei and Christiano were his OpenAI safety colleagues who went on to co-found Anthropic; Hassabis co-founded DeepMind rather than joining it from OpenAI.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "pa4rkh1hu3w0f" }, { "id": "ot-alignment-figures-3", "shape": "mcq", "tags": [ "lecun", "safety-letter" ], "prompt": { "modality": "text", "value": "Which prominent AI researcher did NOT sign the Center for AI Safety's May 2023 one-sentence letter?" }, "options": [ { "modality": "text", "value": "Yann LeCun" }, { "modality": "text", "value": "Yoshua Bengio" }, { "modality": "text", "value": "Sam Altman" }, { "modality": "text", "value": "Geoffrey Hinton" } ], "correctIndex": 0, "explanation": "Yann LeCun declined, consistent with his long-held scepticism about near-term catastrophic risk. Bengio, Hinton and Altman all signed, alongside Hassabis and Amodei.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "izyg7qku3h54" }, { "id": "ot-alignment-figures-4", "shape": "mcq", "tags": [ "safety-letter", "cais" ], "prompt": { "modality": "text", "value": "The May 2023 one-sentence letter compared AI extinction risk to which other societal-scale risks?" }, "options": [ { "modality": "text", "value": "Climate change and famine" }, { "modality": "text", "value": "Pandemics and nuclear war" }, { "modality": "text", "value": "Asteroid impacts and supervolcanoes" }, { "modality": "text", "value": "Cyberattacks and financial collapse" } ], "correctIndex": 1, "explanation": "The single sentence named pandemics and nuclear war as the comparison class. The other pairings are plausible societal risks but were not the ones the letter invoked.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1o36ooo12ajx6e" }, { "id": "ot-alignment-figures-5", "shape": "mcq", "tags": [ "safety-letter", "bengio" ], "prompt": { "modality": "text", "value": "Which group correctly lists researchers named in the guide as signatories of the May 2023 one-sentence letter?" }, "options": [ { "modality": "text", "value": "Hinton, LeCun, Hassabis and Amodei" }, { "modality": "text", "value": "Bengio, LeCun, Altman and Hassabis" }, { "modality": "text", "value": "LeCun, Irving, Buchanan and Sunak" }, { "modality": "text", "value": "Hinton, Bengio, Hassabis, Altman and Amodei" } ], "correctIndex": 3, "explanation": "The guide names Hinton, Bengio, Hassabis, Altman and Amodei among the signatories. Every other option includes Yann LeCun, who pointedly declined to sign.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "hbrvrv4u4del" }, { "id": "ot-govt-response-1", "shape": "mcq", "tags": [ "buchanan", "white-house" ], "prompt": { "modality": "text", "value": "Whom did the Biden administration appoint to its new senior AI coordination role in June 2023?" }, "options": [ { "modality": "text", "value": "Geoffrey Irving" }, { "modality": "text", "value": "Mustafa Suleyman" }, { "modality": "text", "value": "Ben Buchanan" }, { "modality": "text", "value": "Dario Amodei" } ], "correctIndex": 2, "explanation": "Ben Buchanan, a Georgetown professor of AI and cybersecurity working at the National Security Council, took the AI czar role. Suleyman ran Inflection and Amodei Anthropic — labs that signed the commitments.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "u47rk2f6b2c8" }, { "id": "ot-govt-response-2", "shape": "mcq", "tags": [ "buchanan", "background" ], "prompt": { "modality": "text", "value": "What was Ben Buchanan's academic and professional background before becoming AI czar?" }, "options": [ { "modality": "text", "value": "A DeepMind alignment researcher recruited from OpenAI", "short": "DeepMind alignment researcher" }, { "modality": "text", "value": "A semiconductor export-control lawyer at Commerce", "short": "Commerce export-control lawyer" }, { "modality": "text", "value": "A Georgetown professor of AI and cybersecurity on the National Security Council", "short": "Georgetown AI/cyber professor" }, { "modality": "text", "value": "A Center for AI Safety policy director", "short": "Center for AI Safety director" } ], "correctIndex": 2, "explanation": "Buchanan came from Georgetown, specialising in AI and cybersecurity, via the National Security Council. The DeepMind alignment description fits Geoffrey Irving instead.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "pz2blv1d9ecmx" }, { "id": "ot-govt-response-3", "shape": "mcq", "tags": [ "commitments", "white-house" ], "prompt": { "modality": "text", "value": "Where were the July 2023 voluntary AI commitments signed?" }, "options": [ { "modality": "text", "value": "At the Georgetown School of Foreign Service" }, { "modality": "text", "value": "At Bletchley Park" }, { "modality": "text", "value": "At OpenAI's headquarters" }, { "modality": "text", "value": "At the White House" } ], "correctIndex": 3, "explanation": "The 21 July 2023 commitments were signed at the White House. Bletchley Park hosted the separate international summit that November; Georgetown was Buchanan's academic home, not the venue.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "12ba4va15k94g0" }, { "id": "ot-govt-response-4", "shape": "mcq", "tags": [ "commitments", "enforcement" ], "prompt": { "modality": "text", "value": "Which was NOT among the voluntary commitments the labs signed in July 2023?" }, "options": [ { "modality": "text", "value": "Binding financial penalties for backsliding" }, { "modality": "text", "value": "Cybersecurity investment" }, { "modality": "text", "value": "Pre-release safety testing" }, { "modality": "text", "value": "Public reporting of capabilities" } ], "correctIndex": 0, "explanation": "The pledges carried no enforcement mechanism — Buchanan's leverage was reputational. Safety testing, cybersecurity investment and capability reporting were the three actual commitments.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "i27gdj19yqsx5" }, { "id": "ot-govt-response-5", "shape": "mcq", "tags": [ "commitments", "labs" ], "prompt": { "modality": "text", "value": "How did major US frontier labs respond to the July 2023 voluntary commitments?" }, "options": [ { "modality": "text", "value": "Meta and Amazon publicly refused" }, { "modality": "text", "value": "No major US frontier lab declined to sign" }, { "modality": "text", "value": "Only OpenAI and Anthropic signed" }, { "modality": "text", "value": "All declined pending congressional legislation" } ], "correctIndex": 1, "explanation": "Google, OpenAI, Anthropic, Meta, Amazon, Microsoft and Inflection all signed — not a single major US frontier lab declined, which is what made the pledges a coordinated first.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "jyhkwr14d6z5h" }, { "id": "ot-bletchley-1", "shape": "mcq", "tags": [ "bletchley", "numbers" ], "prompt": { "modality": "text", "value": "How many countries sent delegations that signed the Bletchley Declaration?" }, "options": [ { "modality": "text", "value": "28" }, { "modality": "text", "value": "50" }, { "modality": "text", "value": "18" }, { "modality": "text", "value": "38" } ], "correctIndex": 0, "explanation": "Delegates from 28 countries gathered at Bletchley Park on 1 November 2023, and all 28 delegations signed the Declaration.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "db0otl3y5v03" }, { "id": "ot-bletchley-2", "shape": "mcq", "tags": [ "bletchley", "dates" ], "prompt": { "modality": "text", "value": "In what month and year was the first international AI Safety Summit held at Bletchley Park?" }, "options": [ { "modality": "text", "value": "November 2023" }, { "modality": "text", "value": "November 2022" }, { "modality": "text", "value": "July 2023" }, { "modality": "text", "value": "October 2023" } ], "correctIndex": 0, "explanation": "The summit opened 1 November 2023. October 2023 was Biden's executive order and July 2023 the voluntary commitments — the summit came after both.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1q57zoh1soa9ir" }, { "id": "ot-bletchley-3", "shape": "mcq", "tags": [ "bletchley", "turing" ], "prompt": { "modality": "text", "value": "Why was Bletchley Park a pointed venue for the first AI Safety Summit?" }, "options": [ { "modality": "text", "value": "It housed the UK's largest AI supercomputer" }, { "modality": "text", "value": "It was DeepMind's original London laboratory" }, { "modality": "text", "value": "It was Alan Turing's wartime codebreaking home" }, { "modality": "text", "value": "It was where the Transformer paper was written" } ], "correctIndex": 2, "explanation": "Bletchley Park was Alan Turing's wartime codebreaking workplace, which gave the summit its symbolism. DeepMind was founded in London but not at Bletchley.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "6bsr351ge5l7j" }, { "id": "ot-bletchley-4", "shape": "mcq", "tags": [ "bletchley", "hassabis" ], "prompt": { "modality": "text", "value": "Whose conversation with Prime Minister Rishi Sunak, about six months earlier, led to the Bletchley summit?" }, "options": [ { "modality": "text", "value": "Demis Hassabis" }, { "modality": "text", "value": "Sam Altman" }, { "modality": "text", "value": "Geoffrey Hinton" }, { "modality": "text", "value": "Ben Buchanan" } ], "correctIndex": 0, "explanation": "Hassabis pitched Sunak on a global summit roughly six months before Bletchley. Buchanan was driving the parallel US executive-order track, not the UK summit.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "197xkx6rdov3g" }, { "id": "ot-chip-china-1", "shape": "mcq", "tags": [ "chip-ban", "dates" ], "prompt": { "modality": "text", "value": "In what year did the Biden administration impose broad export controls on advanced semiconductors bound for China?" }, "options": [ { "modality": "text", "value": "2017" }, { "modality": "text", "value": "2022" }, { "modality": "text", "value": "2020" }, { "modality": "text", "value": "2023" } ], "correctIndex": 1, "explanation": "The controls came in October 2022. 2023 was the year of the voluntary commitments, the executive order and the Bletchley summit; 2017 was the Ke Jie match.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "cz3msqptxq6" }, { "id": "ot-chip-china-2", "shape": "mcq", "tags": [ "chip-ban", "scope" ], "prompt": { "modality": "text", "value": "What did the Biden administration's October 2022 controls actually restrict?" }, "options": [ { "modality": "text", "value": "Export of trained model weights to China" }, { "modality": "text", "value": "Chinese investment in US AI laboratories" }, { "modality": "text", "value": "Chinese researchers' access to US AI conferences" }, { "modality": "text", "value": "The supply of advanced semiconductors to China" } ], "correctIndex": 3, "explanation": "The controls were a wide-ranging ban on supplying advanced semiconductors to China. They targeted hardware, not capital flows, model weights or researcher access.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "vo7tpw16kzzrm" }, { "id": "ot-chip-china-3", "shape": "mcq", "tags": [ "ke-jie", "alphago" ], "prompt": { "modality": "text", "value": "Which world champion's defeat by AlphaGo was described inside Beijing as a 'Sputnik moment'?" }, "options": [ { "modality": "text", "value": "Garry Kasparov" }, { "modality": "text", "value": "Ke Jie" }, { "modality": "text", "value": "Lee Sedol" }, { "modality": "text", "value": "Fan Hui" } ], "correctIndex": 1, "explanation": "AlphaGo's defeat of Ke Jie, played on a Chinese board, prompted the Sputnik framing in Beijing. Lee Sedol's and Fan Hui's earlier losses, and Kasparov's chess defeat, were not the trigger described here.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1227e9o1q6zfa" }, { "id": "ot-chip-china-4", "shape": "mcq", "tags": [ "sputnik", "dates" ], "prompt": { "modality": "text", "value": "From what year has Hassabis tracked China's competitive intent — the year AlphaGo's win over the reigning Go champion was called a 'Sputnik moment'?" }, "options": [ { "modality": "text", "value": "2017" }, { "modality": "text", "value": "2015" }, { "modality": "text", "value": "2020" }, { "modality": "text", "value": "2022" } ], "correctIndex": 0, "explanation": "The match was in 2017. By 2022 China's State Council already had a formal AI strategy with 2030 targets — the strategy followed the Sputnik moment, it did not precede it.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "luay5lcd5usv" }, { "id": "ot-chip-china-5", "shape": "mcq", "tags": [ "chip-ban", "mallaby" ], "prompt": { "modality": "text", "value": "How does Mallaby read the significance of the chip-export ban?" }, "options": [ { "modality": "text", "value": "It was a symbolic gesture with no effect on the AI race", "short": "Symbolic, no effect on AI race" }, { "modality": "text", "value": "It gave China an unbeatable lead in advanced semiconductors", "short": "Gave China a semiconductor lead" }, { "modality": "text", "value": "It was the direct cause of the Bletchley Declaration", "short": "Caused the Bletchley Declaration" }, { "modality": "text", "value": "It closed the most consequential lane of US-China technology cooperation", "short": "Closed US-China tech cooperation" } ], "correctIndex": 3, "explanation": "Mallaby reads the ban as closing the most consequential lane of US-China cooperation, merging the AI race and the chip war into one story. Bletchley traced to Hassabis and Sunak, not to the ban.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "18aogcmfvao1k" }, { "id": "ot-gemini-define-1", "shape": "mcq", "tags": [ "gemini", "google-deepmind" ], "prompt": { "modality": "text", "value": "What is Gemini?" }, "options": [ { "modality": "text", "value": "Google DeepMind's flagship family of multimodal AI models, launched in 2023", "short": "Google DeepMind multimodal models" }, { "modality": "text", "value": "OpenAI's successor to GPT-4, launched in 2023", "short": "OpenAI's GPT-4 successor model" }, { "modality": "text", "value": "Anthropic's mechanistic-interpretability research programme", "short": "Anthropic interpretability program" }, { "modality": "text", "value": "Meta's open-weights model family released alongside Llama", "short": "Meta open-weights model family" } ], "correctIndex": 0, "explanation": "Gemini is Google DeepMind's flagship multimodal model family, launched in 2023 as the lab's answer to OpenAI's GPT-4 and its return to the research frontier. It is a Google DeepMind product, not an OpenAI one.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "wgzqaj1593uhl" }, { "id": "ot-gemini-define-2", "shape": "mcq", "tags": [ "gemini", "timeline" ], "prompt": { "modality": "text", "value": "On what date did Google DeepMind launch Gemini?" }, "options": [ { "modality": "text", "value": "30 November 2022" }, { "modality": "text", "value": "15 February 2024" }, { "modality": "text", "value": "6 December 2022" }, { "modality": "text", "value": "6 December 2023" } ], "correctIndex": 3, "explanation": "Gemini launched on 6 December 2023, roughly fourteen months after ChatGPT (30 November 2022). 15 February 2024 is the later Gemini 1.5 Pro release.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "4kq0w41n4tpaq" }, { "id": "ot-gemini-define-3", "shape": "mcq", "tags": [ "gemini-ultra", "launch" ], "prompt": { "modality": "text", "value": "At the launch, which Gemini variant was the headline model previewed against GPT-4 rather than shipped immediately?" }, "options": [ { "modality": "text", "value": "Gemini 1.5 Pro" }, { "modality": "text", "value": "Gemini Nano" }, { "modality": "text", "value": "Gemini Flash" }, { "modality": "text", "value": "Gemini Ultra" } ], "correctIndex": 3, "explanation": "Two smaller variants shipped immediately; the headline Gemini Ultra was only previewed, benchmarked against GPT-4 on MMLU. Gemini 1.5 Pro was a separate release in February 2024.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1b9fno11ytu3yv" }, { "id": "ot-gemini-define-4", "shape": "mcq", "tags": [ "gemini", "launch" ], "prompt": { "modality": "text", "value": "Which statement about Gemini's December 2023 launch is NOT true?" }, "options": [ { "modality": "text", "value": "The launch arrived roughly fourteen months after ChatGPT", "short": "Arrived ~14 months after ChatGPT" }, { "modality": "text", "value": "Gemini Ultra was previewed against GPT-4 on the MMLU benchmark", "short": "Ultra previewed vs GPT-4 on MMLU" }, { "modality": "text", "value": "It was the first frontier model built jointly by the merged London and Mountain View teams", "short": "First model from the merged teams" }, { "modality": "text", "value": "Every variant, including the headline model, shipped to users immediately", "short": "Every variant shipped immediately" } ], "correctIndex": 3, "explanation": "Only two smaller variants shipped immediately — Gemini Ultra was previewed, not released. The other three statements match Mallaby's account of the launch.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1ytxow1yv52km" }, { "id": "ot-gemini-define-5", "shape": "mcq", "tags": [ "merger", "gemini" ], "prompt": { "modality": "text", "value": "In Mallaby's account, why did the Gemini launch matter for Google DeepMind beyond the model itself?" }, "options": [ { "modality": "text", "value": "It was the first proof the Brain–DeepMind merger could ship at frontier scale", "short": "First proof the merger could ship" }, { "modality": "text", "value": "It was the first commercial product DeepMind had ever sold", "short": "First product DeepMind ever sold" }, { "modality": "text", "value": "It was the first model trained entirely in London without Mountain View", "short": "First model trained only in London" }, { "modality": "text", "value": "It was the first time Google published its full training data", "short": "First full training-data release" } ], "correctIndex": 0, "explanation": "Gemini was the first test of whether the merger's synergy story was real — late by Silicon Valley standards, but proof the merged organisation could ship at frontier scale. London only set the research agenda later, with Gemini 1.5 Pro.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "144b3ns60f6um" }, { "id": "ot-mmlu-1", "shape": "mcq", "tags": [ "mmlu", "benchmark" ], "prompt": { "modality": "text", "value": "What does the acronym MMLU stand for?" }, "options": [ { "modality": "text", "value": "Multi-Metric Language Uniformity" }, { "modality": "text", "value": "Massive Model Learning Universe" }, { "modality": "text", "value": "Multimodal Machine Learning Utility" }, { "modality": "text", "value": "Massive Multitask Language Understanding" } ], "correctIndex": 3, "explanation": "MMLU is Massive Multitask Language Understanding — the benchmark on which Gemini Ultra was previewed against GPT-4.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "xfaovn11kws11" }, { "id": "ot-mmlu-2", "shape": "mcq", "tags": [ "mmlu", "benchmark" ], "prompt": { "modality": "text", "value": "How many subjects does the MMLU benchmark cover?" }, "options": [ { "modality": "text", "value": "40" }, { "modality": "text", "value": "57" }, { "modality": "text", "value": "24" }, { "modality": "text", "value": "100" } ], "correctIndex": 1, "explanation": "MMLU spans 57 subjects, covering maths, humanities, ethics and law — its breadth is the point, testing general knowledge rather than one narrow skill.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "19de75fypjo79" }, { "id": "ot-mmlu-3", "shape": "mcq", "tags": [ "mmlu", "gemini-ultra" ], "prompt": { "modality": "text", "value": "What was Gemini Ultra's announced MMLU score?" }, "options": [ { "modality": "text", "value": "90%" }, { "modality": "text", "value": "86%" }, { "modality": "text", "value": "44%" }, { "modality": "text", "value": "95%" } ], "correctIndex": 0, "explanation": "Gemini Ultra's announced MMLU score was 90%, edging past GPT-4's published 86%. 44% was GPT-3's 2020 score.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "gx70xbdv9mit" }, { "id": "ot-mmlu-4", "shape": "mcq", "tags": [ "mmlu", "benchmark" ], "prompt": { "modality": "text", "value": "Which of these is NOT one of the areas MMLU's 57 subjects are described as spanning?" }, "options": [ { "modality": "text", "value": "Maths" }, { "modality": "text", "value": "Computer vision" }, { "modality": "text", "value": "Law" }, { "modality": "text", "value": "Humanities" } ], "correctIndex": 1, "explanation": "MMLU spans maths, humanities, ethics and law. Computer vision is not part of the benchmark — MMLU tests language-based multitask knowledge.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "137ndzrubxeat" }, { "id": "ot-mmlu-5", "shape": "mcq", "tags": [ "mmlu", "progress" ], "prompt": { "modality": "text", "value": "Which sequence correctly ranks the MMLU scores for GPT-3 (2020), GPT-4 (2023) and Gemini Ultra?" }, "options": [ { "modality": "text", "value": "60% → 86% → 95%" }, { "modality": "text", "value": "44% → 86% → 90%" }, { "modality": "text", "value": "56% → 80% → 90%" }, { "modality": "text", "value": "44% → 90% → 86%" } ], "correctIndex": 1, "explanation": "GPT-3 scored 44% in 2020, GPT-4 reached 86% in 2023, and Gemini Ultra hit 90% — the first system announced above top human experts. Each bump is roughly a year of student-level progress.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1de1k931pspe7d" }, { "id": "ot-gemini15-1", "shape": "mcq", "tags": [ "gemini-1-5-pro", "context-window" ], "prompt": { "modality": "text", "value": "What was Gemini 1.5 Pro's headline capability at launch?" }, "options": [ { "modality": "text", "value": "Native real-time voice conversation" }, { "modality": "text", "value": "A 90% score on the MMLU benchmark" }, { "modality": "text", "value": "A 32,000-token context window" }, { "modality": "text", "value": "A million-token context window" } ], "correctIndex": 3, "explanation": "Gemini 1.5 Pro's headline capability was a million-token context window, paired with a sparse mixture-of-experts architecture. The 90% MMLU score belonged to Gemini Ultra at the earlier December 2023 launch.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "gj9vog1xu63wq" }, { "id": "ot-gemini15-2", "shape": "mcq", "tags": [ "gemini-1-5-pro", "timeline" ], "prompt": { "modality": "text", "value": "When did Google DeepMind release Gemini 1.5 Pro?" }, "options": [ { "modality": "text", "value": "30 November 2023" }, { "modality": "text", "value": "15 February 2025" }, { "modality": "text", "value": "15 February 2024" }, { "modality": "text", "value": "6 December 2023" } ], "correctIndex": 2, "explanation": "Gemini 1.5 Pro was released on 15 February 2024. 6 December 2023 was the original Gemini launch.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "k4xncu1soma1g" }, { "id": "ot-gemini15-3", "shape": "mcq", "tags": [ "mixture-of-experts", "architecture" ], "prompt": { "modality": "text", "value": "What does a sparse mixture-of-experts architecture mean for how the model runs?" }, "options": [ { "modality": "text", "value": "Only a fraction of the model's parameters activate for any given input", "short": "Few parameters fire per input" }, { "modality": "text", "value": "Human expert raters score every output before it is returned", "short": "Human experts score each output" }, { "modality": "text", "value": "The model is retrained from scratch for each specialist domain", "short": "Retrained per specialist domain" }, { "modality": "text", "value": "Several separate models vote and the majority answer wins", "short": "Models vote; majority answer wins" } ], "correctIndex": 0, "explanation": "In a sparse mixture-of-experts design only a fraction of parameters activate per input, which keeps inference costs manageable despite the model's size. It is not an ensemble vote or a human-review step.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "mqlrn1gu9gn3" }, { "id": "ot-gemini15-4", "shape": "mcq", "tags": [ "rlhf", "post-training" ], "prompt": { "modality": "text", "value": "After David Silver quit the post-training team, which approach actually shipped Gemini's post-training upgrade?" }, "options": [ { "modality": "text", "value": "Silver's machine-RL pipeline, scaled up by a larger team", "short": "Silver's machine-RL pipeline" }, { "modality": "text", "value": "A London strike team using sparse mixture-of-experts", "short": "London team using sparse MoE" }, { "modality": "text", "value": "A Bard veterans team focused on data cleaning and RLHF widgets", "short": "Bard veterans doing data + RLHF" }, { "modality": "text", "value": "Geoffrey Irving's mechanistic-interpretability circuits", "short": "Irving's interpretability circuits" } ], "correctIndex": 2, "explanation": "Silver quit after ~6 months because language quality gives RL no clear reward signal. A Bard veterans team concentrating on data cleaning and RLHF widgets shipped the upgrade instead — not sophisticated machine RL.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "12ty1sg1m7nw8i" }, { "id": "ot-firing-1", "shape": "mcq", "tags": [ "altman", "openai-board" ], "prompt": { "modality": "text", "value": "Which pair does Mallaby identify as the two triggers for the OpenAI board's November 2023 firing of Altman?" }, "options": [ { "modality": "text", "value": "The 100x profit cap, and an undisclosed deal with Microsoft", "short": "Profit cap + secret Microsoft deal" }, { "modality": "text", "value": "A leaked model, and refusing to sign the White House safety commitments", "short": "Leaked model + unsigned safety pledge" }, { "modality": "text", "value": "A failed safety evaluation, and misreporting compute spending", "short": "Failed eval + misreported compute" }, { "modality": "text", "value": "Undisclosed Gulf fundraising, and a pattern of pushing out critical board members", "short": "Gulf fundraising + ousting critics" } ], "correctIndex": 3, "explanation": "Mallaby gives exactly two triggers: fundraising overtures to Gulf investors that Altman had not disclosed to the board, and a pattern of attempting to push out independent board members who criticised him.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "17y3jx5rbwn7" }, { "id": "ot-firing-2", "shape": "mcq", "tags": [ "altman", "fundraising" ], "prompt": { "modality": "text", "value": "Undisclosed fundraising overtures to investors in which region formed one of the two triggers?" }, "options": [ { "modality": "text", "value": "Southeast Asia" }, { "modality": "text", "value": "The Persian Gulf" }, { "modality": "text", "value": "Scandinavia" }, { "modality": "text", "value": "Latin America" } ], "correctIndex": 1, "explanation": "Mallaby cites undisclosed fundraising overtures to investors in the Persian Gulf — conversations Altman had not disclosed to the board.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "f1kpznt91lh5" }, { "id": "ot-firing-3", "shape": "mcq", "tags": [ "altman", "openai-board" ], "prompt": { "modality": "text", "value": "A news report says Altman was fired over a failed safety evaluation. On Mallaby's account, what is wrong with that claim?" }, "options": [ { "modality": "text", "value": "Running safety evaluations was the board's own job, not Altman's", "short": "Evals were the board's job, not his" }, { "modality": "text", "value": "The evaluation did fail, but the board judged the lapse too minor to act on", "short": "Eval failed but lapse too minor" }, { "modality": "text", "value": "Evaluations only began after the firing, so none could have failed", "short": "Evals began only after the firing" }, { "modality": "text", "value": "Neither trigger concerned model testing; both were about undisclosed conduct and board control", "short": "No trigger involved model testing" } ], "correctIndex": 3, "explanation": "Mallaby's two triggers were the undisclosed Gulf fundraising and the pattern of pushing out critical board members — questions of trust and governance, not of any safety evaluation result.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "8l32visow74k" }, { "id": "ot-firing-4", "shape": "mcq", "tags": [ "openai-board", "charter" ], "prompt": { "modality": "text", "value": "The OpenAI board's charter charged it with pursuing what mission — the standard against which it judged Altman?" }, "options": [ { "modality": "text", "value": "\"Open research, openly published\"" }, { "modality": "text", "value": "\"Artificial general intelligence for shareholders\"" }, { "modality": "text", "value": "\"Compute at any cost\"" }, { "modality": "text", "value": "\"Safe AI for the benefit of the world\"" } ], "correctIndex": 3, "explanation": "The board's charter charged it with pursuing \"safe AI for the benefit of the world\"; measured against that mission, it concluded Altman could not be trusted.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "zyvrlpuppo6n" }, { "id": "ot-firing-5", "shape": "mcq", "tags": [ "altman", "openai-board" ], "prompt": { "modality": "text", "value": "What does Mallaby say the board underestimated when it fired Altman?" }, "options": [ { "modality": "text", "value": "How thoroughly Silicon Valley and OpenAI's workforce would close ranks behind him", "short": "Staff and Valley rallying to Altman" }, { "modality": "text", "value": "How far ahead Google DeepMind's Gemini already was", "short": "How far ahead Gemini already was" }, { "modality": "text", "value": "How much regulators would intervene in the dispute", "short": "Regulators intervening in the feud" }, { "modality": "text", "value": "How quickly Microsoft would withdraw its funding", "short": "Microsoft quickly pulling funding" } ], "correctIndex": 0, "explanation": "The board underestimated how completely Silicon Valley and OpenAI's own workforce would rally behind Altman within days — the reaction that undid the firing.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "18yorz0ct75zi" }, { "id": "ot-closing-1", "shape": "mcq", "tags": [ "hassabis", "closing" ], "prompt": { "modality": "text", "value": "As Mallaby closes the book, what is Hassabis's dual status?" }, "options": [ { "modality": "text", "value": "A Nobel laureate and chief executive of one of the most powerful AI labs", "short": "Nobel laureate and AI-lab CEO" }, { "modality": "text", "value": "A Turing Award winner and a UK government minister", "short": "Turing winner and UK minister" }, { "modality": "text", "value": "A chess grandmaster and OpenAI board member", "short": "Chess grandmaster and OpenAI board" }, { "modality": "text", "value": "A Nobel laureate and chief scientist of the AI Safety Institute", "short": "Nobel laureate and AISI scientist" } ], "correctIndex": 0, "explanation": "Mallaby ends with Hassabis as simultaneously a Nobel laureate and the CEO of one of the most powerful AI labs in the world. It was Geoffrey Irving who became chief scientist at the UK's AI Safety Institute.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "et130c1t5g4yi" }, { "id": "ot-closing-2", "shape": "mcq", "tags": [ "closing", "paradox" ], "prompt": { "modality": "text", "value": "What paradox does Mallaby close the book on, and decline to resolve?" }, "options": [ { "modality": "text", "value": "That larger models became cheaper to run than smaller ones", "short": "Big models cheaper than small ones" }, { "modality": "text", "value": "That AI safety research is funded mostly by the labs it scrutinises", "short": "Safety research funded by the labs" }, { "modality": "text", "value": "That the most morally serious figure in the race is also racing hardest to build the technology he warns about", "short": "Warns of AI yet races to build it" }, { "modality": "text", "value": "That the UK produced the science while the US captured the profits", "short": "UK made the science, US took profits" } ], "correctIndex": 2, "explanation": "Mallaby's closing paradox is that the most morally serious major figure in the race is also racing as hard as anybody to build the technology he has spent his career warning the world about — and he does not pretend to resolve it.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1ht8owzvup22p" }, { "id": "ot-closing-3", "shape": "mcq", "tags": [ "hassabis", "closing" ], "prompt": { "modality": "text", "value": "Which of these is NOT part of how Mallaby closes the book on Hassabis?" }, "options": [ { "modality": "text", "value": "His preference for \"realist\" over \"doomer\" or \"accelerationist\"", "short": "Prefers 'realist' label" }, { "modality": "text", "value": "His decision to step back from the lab to work full-time on safety", "short": "Stepped back for AI safety" }, { "modality": "text", "value": "His standing as a Nobel laureate", "short": "Nobel laureate standing" }, { "modality": "text", "value": "His job running one of the most powerful AI labs in the world", "short": "Runs a powerful AI lab" } ], "correctIndex": 1, "explanation": "Mallaby's closing register holds the two roles together — Nobel laureate and sitting CEO — alongside the \"realist\" label. Hassabis never steps back from the lab; that tension is the point of the ending.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1jvjx7xph1v8b" }, { "id": "ot-closing-4", "shape": "mcq", "tags": [ "closing", "mallaby" ], "prompt": { "modality": "text", "value": "A reader says Mallaby ends by judging Hassabis a hypocrite. Why does that misread the closing register?" }, "options": [ { "modality": "text", "value": "Mallaby settles the question by crediting the Nobel Prize", "short": "Settles it via the Nobel Prize" }, { "modality": "text", "value": "Mallaby explicitly declines to resolve the paradox he closes on", "short": "Declines to resolve the paradox" }, { "modality": "text", "value": "Mallaby concludes Hassabis is at heart a doomer", "short": "Concludes Hassabis is a doomer" }, { "modality": "text", "value": "Mallaby argues the tension dissolved once Gemini 1.5 Pro shipped", "short": "Says Gemini 1.5 Pro dissolved it" } ], "correctIndex": 1, "explanation": "Mallaby does not pretend to resolve the paradox — he leaves the two Hassabises standing side by side. He offers no verdict, and the \"realist\" label is Hassabis's own, not Mallaby's judgement.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "zuzxrf1a5rqgd" }, { "id": "ot-trio-1", "shape": "mcq", "tags": [ "cofounders", "agi" ], "prompt": { "modality": "text", "value": "Which DeepMind cofounder coined the term \"AGI\" with Marcus Hutter in 2002?" }, "options": [ { "modality": "text", "value": "Shane Legg" }, { "modality": "text", "value": "Mustafa Suleyman" }, { "modality": "text", "value": "Demis Hassabis" }, { "modality": "text", "value": "David Silver" } ], "correctIndex": 0, "explanation": "Shane Legg, the cerebral New Zealander of the three cofounders, coined \"AGI\" with Marcus Hutter in 2002 and later gave the Halloween Scenario lecture that drew Hassabis in.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "utnycujlsmcw" }, { "id": "ot-trio-2", "shape": "mcq", "tags": [ "cofounders", "microsoft" ], "prompt": { "modality": "text", "value": "Which DeepMind cofounder was running Microsoft AI by 2024?" }, "options": [ { "modality": "text", "value": "Demis Hassabis" }, { "modality": "text", "value": "Mustafa Suleyman" }, { "modality": "text", "value": "Ilya Sutskever" }, { "modality": "text", "value": "Shane Legg" } ], "correctIndex": 1, "explanation": "Suleyman was the operator of the three cofounders and ended up running Microsoft AI by 2024. Hassabis stayed on as DeepMind's CEO; Sutskever was OpenAI's chief scientist, not a DeepMind cofounder.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "tm8ko0m99uza" }, { "id": "ot-trio-3", "shape": "mcq", "tags": [ "cofounders", "nobel" ], "prompt": { "modality": "text", "value": "Which DeepMind cofounder shared the 2024 Nobel Prize in Chemistry?" }, "options": [ { "modality": "text", "value": "David Silver" }, { "modality": "text", "value": "Shane Legg" }, { "modality": "text", "value": "Mustafa Suleyman" }, { "modality": "text", "value": "Demis Hassabis" } ], "correctIndex": 3, "explanation": "Hassabis — the chess prodigy, neuroscientist, cofounder and CEO — shared the 2024 Nobel Prize in Chemistry with John Jumper for the AlphaFold work.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "j16b301y80y9i" }, { "id": "ot-trio-4", "shape": "mcq", "tags": [ "cofounders" ], "prompt": { "modality": "text", "value": "Which of these people was NOT one of DeepMind's three cofounders?" }, "options": [ { "modality": "text", "value": "Shane Legg" }, { "modality": "text", "value": "Mustafa Suleyman" }, { "modality": "text", "value": "Demis Hassabis" }, { "modality": "text", "value": "David Silver" } ], "correctIndex": 3, "explanation": "The founding trio was Hassabis, Legg, and Suleyman. Silver was Hassabis's Cambridge friend and the architect of AlphaGo — a long-running technical second, but not a cofounder.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1xo9wj01tuhjny" }, { "id": "ot-trio-5", "shape": "mcq", "tags": [ "agi", "dates" ], "prompt": { "modality": "text", "value": "In what year did Shane Legg and Marcus Hutter coin the term \"AGI\"?" }, "options": [ { "modality": "text", "value": "1997" }, { "modality": "text", "value": "2010" }, { "modality": "text", "value": "2002" }, { "modality": "text", "value": "2014" } ], "correctIndex": 2, "explanation": "Legg and Hutter coined \"AGI\" in 2002. 2010 was the Singularity Summit pitch to Thiel and 2014 was the Google acquisition — later milestones, not the coinage.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1d1xk0cp82knm" }, { "id": "ot-sci-1", "shape": "mcq", "tags": [ "atari-dqn", "scientists" ], "prompt": { "modality": "text", "value": "Who led the Atari DQN work — the foundational reinforcement-learning system every later DeepMind result built on?" }, "options": [ { "modality": "text", "value": "John Jumper" }, { "modality": "text", "value": "Geoffrey Irving" }, { "modality": "text", "value": "Vlad Mnih" }, { "modality": "text", "value": "David Silver" } ], "correctIndex": 2, "explanation": "Vlad Mnih led the Atari DQN work, the reinforcement-learning foundation the book says every later DeepMind result stood on. Irving's DeepMind work was the alignment programme, not a game-playing system.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "r7q7pi8eur84" }, { "id": "ot-sci-2", "shape": "mcq", "tags": [ "scientists", "silver" ], "prompt": { "modality": "text", "value": "Which scientist does the book describe as Hassabis's Cambridge friend and a long-running technical second at DeepMind?" }, "options": [ { "modality": "text", "value": "Geoffrey Irving" }, { "modality": "text", "value": "David Silver" }, { "modality": "text", "value": "Vlad Mnih" }, { "modality": "text", "value": "John Jumper" } ], "correctIndex": 1, "explanation": "David Silver was Hassabis's friend from Cambridge and DeepMind's long-running technical second. Jumper arrived only in October 2017, and Mnih is never described as a Cambridge connection.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "51cdcgqcc7sa" }, { "id": "ot-sci-3", "shape": "mcq", "tags": [ "alphafold", "scientists" ], "prompt": { "modality": "text", "value": "Which scientist arrived at DeepMind in October 2017 and went on to lead AlphaFold 2?" }, "options": [ { "modality": "text", "value": "John Jumper" }, { "modality": "text", "value": "Geoffrey Irving" }, { "modality": "text", "value": "Vlad Mnih" }, { "modality": "text", "value": "David Silver" } ], "correctIndex": 0, "explanation": "John Jumper joined DeepMind in October 2017 and led AlphaFold 2. Irving's DeepMind work was the alignment programme, not a protein-folding system.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "c2st207qsxwy" }, { "id": "ot-sci-4", "shape": "mcq", "tags": [ "scientists", "projects" ], "prompt": { "modality": "text", "value": "Which scientist–project pairing is NOT correct?" }, "options": [ { "modality": "text", "value": "Vlad Mnih — AlphaFold 2" }, { "modality": "text", "value": "David Silver — AlphaGo" }, { "modality": "text", "value": "John Jumper — AlphaFold 2" }, { "modality": "text", "value": "Vlad Mnih — Atari DQN" } ], "correctIndex": 0, "explanation": "AlphaFold 2 was led by John Jumper; Mnih's contribution was the Atari DQN. The other three pairings are exactly as the book gives them.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1ip0gdtvl8sln" }, { "id": "ot-sci-5", "shape": "mcq", "tags": [ "alphafold", "nobel" ], "prompt": { "modality": "text", "value": "John Jumper shared the 2024 Nobel Prize in Chemistry with which DeepMind figure?" }, "options": [ { "modality": "text", "value": "Demis Hassabis" }, { "modality": "text", "value": "David Silver" }, { "modality": "text", "value": "Shane Legg" }, { "modality": "text", "value": "Vlad Mnih" } ], "correctIndex": 0, "explanation": "Jumper shared the 2024 Nobel Prize in Chemistry with Demis Hassabis for the AlphaFold work. Silver, Legg, and Mnih were central to DeepMind but were not laureates.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "37v32ag6klno" }, { "id": "ot-money-1", "shape": "mcq", "tags": [ "thiel", "funding" ], "prompt": { "modality": "text", "value": "Who was DeepMind's first investor, anchoring the round after the 2010 Singularity Summit pitch and muttering \"Demis's destiny\" walking to the car?" }, "options": [ { "modality": "text", "value": "Peter Thiel" }, { "modality": "text", "value": "Sam Altman" }, { "modality": "text", "value": "Larry Page" }, { "modality": "text", "value": "Sergey Brin" } ], "correctIndex": 0, "explanation": "Peter Thiel, through Founders Fund, anchored DeepMind's first round after the 2010 Singularity Summit pitch and was the source of the \"Demis's destiny\" line. Altman's role in the book is at OpenAI.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "179bd77kotk7t" }, { "id": "ot-money-2", "shape": "mcq", "tags": [ "google", "acquisition" ], "prompt": { "modality": "text", "value": "Which pair brought DeepMind into Google in 2014?" }, "options": [ { "modality": "text", "value": "Larry Page and Sergey Brin" }, { "modality": "text", "value": "Peter Thiel and Marcus Hutter" }, { "modality": "text", "value": "Eric Schmidt and Sundar Pichai" }, { "modality": "text", "value": "Sam Altman and Ilya Sutskever" } ], "correctIndex": 0, "explanation": "Google cofounders Larry Page and Sergey Brin acquired DeepMind in 2014. Thiel was the first investor, not the acquirer; Altman and Sutskever were OpenAI's protagonists.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "ljgsmicz3xhg" }, { "id": "ot-money-3", "shape": "mcq", "tags": [ "altman", "openai" ], "prompt": { "modality": "text", "value": "On what date was OpenAI CEO Sam Altman fired — the firing Mallaby treats as the death of any institutional brake on the race?" }, "options": [ { "modality": "text", "value": "17 November 2023" }, { "modality": "text", "value": "1 October 2023" }, { "modality": "text", "value": "30 November 2022" }, { "modality": "text", "value": "22 November 2023" } ], "correctIndex": 0, "explanation": "Altman was fired on 17 November 2023 and reinstated five days later, around 22 November — the reinstatement date, not the firing date, is the tempting distractor.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "d8low816g7req" }, { "id": "ot-money-4", "shape": "mcq", "tags": [ "sutskever", "openai" ], "prompt": { "modality": "text", "value": "Ilya Sutskever, OpenAI's chief scientist, engineered the board revolt — then reversed himself when what happened?" }, "options": [ { "modality": "text", "value": "Founders Fund threatened to pull its investment" }, { "modality": "text", "value": "The board reinstated Altman on its own" }, { "modality": "text", "value": "Google offered to buy OpenAI outright" }, { "modality": "text", "value": "Microsoft offered to absorb the company" } ], "correctIndex": 3, "explanation": "Sutskever reversed himself once Microsoft offered to absorb OpenAI. Altman's reinstatement followed that reversal rather than causing it, and no Founders Fund or Google threat is part of the account.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "y8lpx84bnpaa" }, { "id": "ot-safety-1", "shape": "mcq", "tags": [ "irving", "safety" ], "prompt": { "modality": "text", "value": "Who built DeepMind's alignment programme?" }, "options": [ { "modality": "text", "value": "Ben Buchanan" }, { "modality": "text", "value": "Geoffrey Irving" }, { "modality": "text", "value": "David Silver" }, { "modality": "text", "value": "Shane Legg" } ], "correctIndex": 1, "explanation": "Geoffrey Irving built DeepMind's alignment programme. Silver's work was AlphaGo and Legg's was the AGI coinage — neither ran DeepMind's safety research.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1ly1da8qx8p8" }, { "id": "ot-safety-2", "shape": "mcq", "tags": [ "buchanan", "policy" ], "prompt": { "modality": "text", "value": "Who served as the Biden White House's AI czar?" }, "options": [ { "modality": "text", "value": "Geoffrey Irving" }, { "modality": "text", "value": "Peter Thiel" }, { "modality": "text", "value": "Mustafa Suleyman" }, { "modality": "text", "value": "Ben Buchanan" } ], "correctIndex": 3, "explanation": "Ben Buchanan was the Biden White House's AI czar from June 2023. Irving's safety work was at a lab and then a national institute, never inside the White House.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "pe0ynq17e2g5k" }, { "id": "ot-safety-3", "shape": "mcq", "tags": [ "bletchley", "summit" ], "prompt": { "modality": "text", "value": "How many countries signed the first AI safety declaration at the November 2023 global AI Safety Summit?" }, "options": [ { "modality": "text", "value": "Twenty-eight" }, { "modality": "text", "value": "One hundred and ninety-five" }, { "modality": "text", "value": "Forty" }, { "modality": "text", "value": "Twelve" } ], "correctIndex": 0, "explanation": "Twenty-eight countries signed the first AI safety declaration in November 2023 — a broad but far-from-universal group, so the UN-wide figure of 195 is wrong.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "q9kjxi2gzyrg" }, { "id": "ot-safety-4", "shape": "mcq", "tags": [ "bletchley", "summit" ], "prompt": { "modality": "text", "value": "The first global AI Safety Summit, in November 2023, was held at which site?" }, "options": [ { "modality": "text", "value": "The Royal Society, London" }, { "modality": "text", "value": "King's College, Cambridge" }, { "modality": "text", "value": "Chequers" }, { "modality": "text", "value": "Bletchley Park" } ], "correctIndex": 3, "explanation": "Bletchley Park — Alan Turing's wartime workplace — hosted the November 2023 summit and gave the declaration signed there its name.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "ysber9jjuh8z" }, { "id": "ot-safety-5", "shape": "mcq", "tags": [ "irving", "buchanan", "policy" ], "prompt": { "modality": "text", "value": "Which figure–role pairing is NOT correct?" }, "options": [ { "modality": "text", "value": "Geoffrey Irving — October 2023 executive order" }, { "modality": "text", "value": "Ben Buchanan — July 2023 voluntary commitments" }, { "modality": "text", "value": "Geoffrey Irving — UK AI Safety Institute" }, { "modality": "text", "value": "Ben Buchanan — Biden White House AI czar" } ], "correctIndex": 0, "explanation": "The October 2023 executive order was Buchanan's work, alongside the July 2023 voluntary commitments. Irving's path ran from DeepMind's alignment programme to the UK AI Safety Institute, not into US policy drafting.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1156ke3mek8md" }, { "id": "ot-key-people-1", "shape": "mcq", "tags": [ "people", "hassabis" ], "prompt": { "modality": "text", "value": "Which figure from the book was a chess prodigy who had mastered the game by age 13?" }, "options": [ { "modality": "text", "value": "Vlad Mnih" }, { "modality": "text", "value": "Shane Legg" }, { "modality": "text", "value": "Demis Hassabis" }, { "modality": "text", "value": "David Silver" } ], "correctIndex": 2, "explanation": "Demis Hassabis was a chess prodigy who mastered the game by 13, and later received the 2024 Nobel Prize in Chemistry for his work on AlphaFold. Silver's game milestone was Go, achieved by a project he led — not personal play.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "q3g9taukbtzg" }, { "id": "ot-key-people-2", "shape": "mcq", "tags": [ "people", "legg", "agi" ], "prompt": { "modality": "text", "value": "Who proposed the concept of Artificial General Intelligence during discussions in 2002, coining the term?" }, "options": [ { "modality": "text", "value": "Mustafa Suleyman" }, { "modality": "text", "value": "John Jumper" }, { "modality": "text", "value": "Demis Hassabis" }, { "modality": "text", "value": "Shane Legg" } ], "correctIndex": 3, "explanation": "Shane Legg, who co-founded the London lab, proposed the concept of Artificial General Intelligence in 2002 discussions. Hassabis co-founded the lab too, but the term itself traces to Legg.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1g4v7gex2cfzw" }, { "id": "ot-key-people-3", "shape": "mcq", "tags": [ "people", "mnih", "atari" ], "prompt": { "modality": "text", "value": "Who led the team that created a neural network to play classic Atari video games, contributing a major AI gaming breakthrough?" }, "options": [ { "modality": "text", "value": "Mustafa Suleyman" }, { "modality": "text", "value": "David Silver" }, { "modality": "text", "value": "Vlad Mnih" }, { "modality": "text", "value": "John Jumper" } ], "correctIndex": 2, "explanation": "Vlad Mnih is recognised for pioneering reinforcement-learning work and led the team behind the network that played classic Atari games. David Silver led the Go milestone instead.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1bw4b0c1jfcu1y" }, { "id": "ot-key-people-4", "shape": "mcq", "tags": [ "people", "roles" ], "prompt": { "modality": "text", "value": "Which person-and-contribution pairing is NOT correct?" }, "options": [ { "modality": "text", "value": "Vlad Mnih — pioneering work in reinforcement learning", "short": "Vlad Mnih — reinforcement learning" }, { "modality": "text", "value": "David Silver — led a project that hit a major milestone in the game Go", "short": "David Silver — Go milestone" }, { "modality": "text", "value": "Shane Legg — pitched a billionaire investor for the lab's initial funding", "short": "Shane Legg — pitched for funding" }, { "modality": "text", "value": "John Jumper — led the London lab's leap into structural biology after 2017", "short": "John Jumper — structural biology" } ], "correctIndex": 2, "explanation": "Pitching the billionaire investor for initial funding was Mustafa Suleyman, not Shane Legg. Silver (Go), Jumper (structural biology, recruited 2017) and Mnih (reinforcement learning) are all correctly paired.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "qyjtx1540nfz" }, { "id": "ot-key-techniques-1", "shape": "mcq", "tags": [ "techniques", "reinforcement-learning" ], "prompt": { "modality": "text", "value": "Which technique has agents learn by interacting with an environment, taking suitable actions to maximize cumulative reward?" }, "options": [ { "modality": "text", "value": "Neural networks" }, { "modality": "text", "value": "Reinforcement learning" }, { "modality": "text", "value": "Deep learning" }, { "modality": "text", "value": "Artificial General Intelligence" } ], "correctIndex": 1, "explanation": "Reinforcement learning is defined by agents interacting with an environment and choosing actions that maximize cumulative reward — the paradigm crucial to DeepMind's early successes. Deep learning is about many-layered networks, not reward-driven action.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1uibfeanvj2jg" }, { "id": "ot-key-techniques-2", "shape": "mcq", "tags": [ "techniques", "deep-learning" ], "prompt": { "modality": "text", "value": "Deep learning is best described as which of the following?" }, "options": [ { "modality": "text", "value": "A hand-built expert system that encodes human rules explicitly", "short": "Hand-coded rule-based system" }, { "modality": "text", "value": "A method where agents learn purely from environment rewards", "short": "Agents learn from rewards" }, { "modality": "text", "value": "A subset of machine learning using many-layered networks inspired by the brain", "short": "Many-layered brain-like nets" }, { "modality": "text", "value": "A theoretical AI that applies intelligence across any task", "short": "General AI for any task" } ], "correctIndex": 2, "explanation": "Deep learning is a subset of machine learning inspired by the brain's structure, using neural networks with many layers to analyze data. The reward-driven description belongs to reinforcement learning; the cross-task description is AGI.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "11uumz413jtzci" }, { "id": "ot-key-techniques-3", "shape": "mcq", "tags": [ "techniques", "neural-network" ], "prompt": { "modality": "text", "value": "A neural network is built from interconnected units that work together to solve complex problems. What are those units called?" }, "options": [ { "modality": "text", "value": "Nodes, or 'neurons'" }, { "modality": "text", "value": "Weights" }, { "modality": "text", "value": "Connections" }, { "modality": "text", "value": "Layers" } ], "correctIndex": 0, "explanation": "A neural network is a computational model inspired by biological brains, consisting of interconnected nodes or 'neurons'. Agents, rewards and environments are the vocabulary of reinforcement learning, not the network's building blocks.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "ylqzabj7qzel" }, { "id": "ot-key-techniques-4", "shape": "mcq", "tags": [ "techniques", "agi" ], "prompt": { "modality": "text", "value": "Which concept names a theoretical AI that can understand, learn, and apply intelligence across a wide range of tasks, replicating human cognitive abilities?" }, "options": [ { "modality": "text", "value": "Reinforcement learning" }, { "modality": "text", "value": "Artificial General Intelligence" }, { "modality": "text", "value": "Deep Q-Network" }, { "modality": "text", "value": "Deep learning" } ], "correctIndex": 1, "explanation": "AGI (Artificial General Intelligence) is the theoretical form of AI that generalises across a wide range of tasks and aims to replicate human cognitive abilities; DeepMind's mission is associated with pursuing it. DQN is a specific algorithm, not a general capability.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "gry0n2fiek38" }, { "id": "ot-key-techniques-5", "shape": "mcq", "tags": [ "techniques", "apply" ], "prompt": { "modality": "text", "value": "A system learns a game by trying actions and improving from a score signal, while a many-layered network reads the screen. Which two techniques does this combine?" }, "options": [ { "modality": "text", "value": "Supervised learning and Monte Carlo tree search" }, { "modality": "text", "value": "Supervised learning and deep learning" }, { "modality": "text", "value": "Reinforcement learning and deep learning" }, { "modality": "text", "value": "Reinforcement learning and Monte Carlo tree search" } ], "correctIndex": 2, "explanation": "Learning from a reward signal is reinforcement learning; the many-layered network reading raw input is deep learning. AGI is a theoretical goal, not a technique that could be applied here.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "b5pv9gel56gu" }, { "id": "ot-key-systems-1", "shape": "mcq", "tags": [ "systems", "alphago" ], "prompt": { "modality": "text", "value": "Which DeepMind program famously defeated world champion Lee Sedol in 2016?" }, "options": [ { "modality": "text", "value": "AlphaGo" }, { "modality": "text", "value": "AlphaZero" }, { "modality": "text", "value": "DQN (Deep Q-Network)" }, { "modality": "text", "value": "AlphaFold" } ], "correctIndex": 0, "explanation": "AlphaGo, built to play Go, beat world champion Lee Sedol in 2016 — a landmark showing reinforcement learning and deep learning combined. AlphaFold predicts protein structures and DQN played Atari games.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "jevqsil8zugg" }, { "id": "ot-key-systems-2", "shape": "mcq", "tags": [ "systems", "alphafold" ], "prompt": { "modality": "text", "value": "Which DeepMind system predicts protein structures, with significant implications for drug discovery and biology?" }, "options": [ { "modality": "text", "value": "AlphaGo" }, { "modality": "text", "value": "DQN (Deep Q-Network)" }, { "modality": "text", "value": "AlphaFold" }, { "modality": "text", "value": "AlphaZero" } ], "correctIndex": 2, "explanation": "AlphaFold uses deep learning to predict protein structures, a breakthrough for drug discovery and biology; it earned Hassabis and Jumper a share of the 2024 Nobel Prize in Chemistry. AlphaGo and DQN play games.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1h9y0medopfw8" }, { "id": "ot-key-systems-3", "shape": "mcq", "tags": [ "systems", "dqn" ], "prompt": { "modality": "text", "value": "DeepMind's Atari-playing system paired deep learning with which reinforcement learning algorithm?" }, "options": [ { "modality": "text", "value": "Backpropagation" }, { "modality": "text", "value": "Monte Carlo tree search" }, { "modality": "text", "value": "Q-learning" }, { "modality": "text", "value": "Gradient descent" } ], "correctIndex": 2, "explanation": "The Atari system combined Q-learning with deep learning techniques to learn from raw pixels. Backpropagation and gradient descent are training mechanics, not reinforcement learning algorithms, and tree search was not what it relied on.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "szmb7y1kcx1ps" }, { "id": "ot-key-systems-4", "shape": "mcq", "tags": [ "systems", "compare" ], "prompt": { "modality": "text", "value": "Which system-and-domain pairing is NOT correct?" }, "options": [ { "modality": "text", "value": "AlphaFold — protein structure prediction for biology and drug discovery", "short": "AlphaFold — protein prediction" }, { "modality": "text", "value": "AlphaGo — the board game Go", "short": "AlphaGo — the board game Go" }, { "modality": "text", "value": "AlphaGo — predicting the three-dimensional structures of proteins", "short": "AlphaGo — protein structures" }, { "modality": "text", "value": "DQN — classic Atari arcade games", "short": "DQN — Atari arcade games" } ], "correctIndex": 2, "explanation": "AlphaGo played Go; protein structure prediction was AlphaFold's domain. The other three pairings match the systems as described.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "12tsa8gkju0va" }, { "id": "ot-key-events-1", "shape": "mcq", "tags": [ "events", "pong" ], "prompt": { "modality": "text", "value": "On 7 July 2013, during a major tennis event, a neural network learned a classic arcade game from scratch and by evening beat the built-in opponent. Which game?" }, "options": [ { "modality": "text", "value": "Pong" }, { "modality": "text", "value": "Asteroids" }, { "modality": "text", "value": "Space Invaders" }, { "modality": "text", "value": "Breakout" } ], "correctIndex": 0, "explanation": "The 7 July 2013 milestone was the day Pong clicked: the network learned the game from scratch and by that evening was beating the built-in opponent, moving toward human-level play.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "10ksw9e1hp4y4g" }, { "id": "ot-key-events-2", "shape": "mcq", "tags": [ "events", "singularity-summit" ], "prompt": { "modality": "text", "value": "In August 2010, a trio of researchers pitched a lab concept aimed at building AGI safely at a conference on the future of technology. In which city was it held?" }, "options": [ { "modality": "text", "value": "Palo Alto" }, { "modality": "text", "value": "San Francisco" }, { "modality": "text", "value": "New York" }, { "modality": "text", "value": "London" } ], "correctIndex": 1, "explanation": "The Singularity Summit took place in San Francisco in August 2010, where the trio pitched their AGI-safety lab. London is where the lab itself was later founded — a tempting mix-up.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "2l1ta31s0hiip" }, { "id": "ot-key-events-3", "shape": "mcq", "tags": [ "events", "atari" ], "prompt": { "modality": "text", "value": "The 2013 'Atari strike team' at the London lab set out to build what?" }, "options": [ { "modality": "text", "value": "A network trained on recordings of human expert playthroughs", "short": "Net trained on human replays" }, { "modality": "text", "value": "An emulator for testing classic arcade hardware", "short": "An arcade-hardware emulator" }, { "modality": "text", "value": "A single neural network that could learn to play multiple games", "short": "One net to learn many games" }, { "modality": "text", "value": "A separate hand-coded program tuned for each individual game", "short": "Hand-coded program per game" } ], "correctIndex": 2, "explanation": "The small 2013 team aimed at one neural network that could learn many classic games — the lab's first major public success in reinforcement learning. Hand-coding per game or copying human demos was exactly what they avoided.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1mlq4wj16k5439" }, { "id": "ot-key-events-4", "shape": "mcq", "tags": [ "events", "musk-page" ], "prompt": { "modality": "text", "value": "At an October 2012 dinner at a tech mogul's home, one guest warned about the dangers of advanced AI while the host disagreed. What is this moment seen as?" }, "options": [ { "modality": "text", "value": "The moment a key alliance in AI development began to fracture", "short": "A key AI alliance fracturing" }, { "modality": "text", "value": "The point when Google agreed to acquire DeepMind", "short": "Google agreeing to buy DeepMind" }, { "modality": "text", "value": "The origin of the independent ethics board condition", "short": "Origin of the ethics-board term" }, { "modality": "text", "value": "The founding of the London AI lab's research team", "short": "Founding the lab's research team" } ], "correctIndex": 0, "explanation": "The Musk–Page exchange on the lawn is seen as the moment a key AI alliance began to fracture. The ethics board came with Google's January 2014 acquisition, a separate event.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "43a6e4taq3y" }, { "id": "ot-agidef-1", "shape": "mcq", "tags": [ "agi", "definition", "infinity-machine" ], "prompt": { "modality": "text", "value": "Mallaby's 'infinity machine' labels AGI. As the book defines it, such a system is capable in principle of doing what?" }, "options": [ { "modality": "text", "value": "Mastering one narrow task better than any human" }, { "modality": "text", "value": "Solving any problem a human mind can frame" }, { "modality": "text", "value": "Predicting protein structures from amino acids" }, { "modality": "text", "value": "Winning closed games like Go and StarCraft" } ], "correctIndex": 1, "explanation": "The guide defines the infinity machine (AGI) as a system capable, in principle, of solving any problem a human mind can frame — the full cognitive range, not one narrow task. The distractors are narrow AI, AlphaFold, and game-playing proving grounds.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1bxy1mi13cxk8o" }, { "id": "fi-ot-book-basics-1", "tags": [ "author", "attribution" ], "title": "Written by Sebastian Mallaby", "body": "Sebastian Mallaby, previously known for writing about hedge funds and venture capital, is the author of The Infinity Machine.", "shape": "fact", "illustration": { "imageSearchTerm": "author writing study portrait", "imagePrompt": "A middle-aged author seated in a book-lined study, writing thoughtfully at a wooden desk.", "alt": "Written by Sebastian Mallaby", "depictable": true, "credit": "Pexels · Elegant black and white portrait of a woman writing with focus and concentration.", "creditUrl": "https://www.pexels.com/photo/black-and-white-portrait-of-woman-writing-35899959/", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-ot-book-basics-1.webp" }, "uid": "qfdc8218rqas0" }, { "id": "fi-ot-book-basics-2", "tags": [ "subject", "hassabis" ], "title": "Hassabis, the book's subject", "body": "The Infinity Machine centers on Demis Hassabis — chess prodigy, game designer, neuroscience PhD, and Nobel laureate.", "shape": "fact", "illustration": { "imageSearchTerm": "Demis Hassabis portrait", "imagePrompt": "A professional portrait of a technology and science leader, confident and thoughtful expression.", "alt": "Hassabis, the book's subject", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Demis_Roussos_in_Kiev.png", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-ot-book-basics-2.webp" }, "uid": "vqb19edthlho" }, { "id": "fi-mcq-publisher", "tags": [ "intro", "vocabulary" ], "title": "Penguin Press, 2026", "body": "The Infinity Machine was published by Penguin Press in 2026.", "shape": "fact", "illustration": { "imageSearchTerm": "hardcover books publishing house", "imagePrompt": "A stack of freshly bound hardcover books with plain unmarked covers, resting on a wooden table.", "alt": "Penguin Press, 2026", "depictable": false, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Rules_of_Play-_Game_Design_Fundamentals_(The_MIT_Press).jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-mcq-publisher.webp" }, "uid": "rgy7521jec910" }, { "id": "fi-mcq-frame", "tags": [ "intro", "agi" ], "title": "DeepMind's AGI-first mission", "body": "Mallaby treats DeepMind's founding mission as a research lab pursuing AGI as a primary goal — a framing built into the company's earliest pitches and incorporation documents from 2010.", "shape": "fact", "illustration": { "imageSearchTerm": "AGI founding mission concept", "imagePrompt": "An abstract glowing neural network forming the shape of a rising path, symbolizing an ambitious founding mission.", "alt": "DeepMind's AGI-first mission", "depictable": false, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Agi_Mishol.jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-mcq-frame.webp" }, "uid": "2nucm41is469e" }, { "id": "fi-ot-thesis-3", "tags": [ "thesis", "hassabis" ], "title": "Hassabis can't slow the race", "body": "Mallaby writes that Hassabis doesn't believe any single actor can slow down the AI race — he sees himself as a participant in it, not someone positioned to act as a brake.", "shape": "fact", "illustration": { "imageSearchTerm": "AI race momentum concept", "imagePrompt": "An abstract single figure running alongside a blurred, fast-moving crowd, unable to slow the momentum.", "alt": "Hassabis can't slow the race", "depictable": false, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Edmund_Blair_Leighton_-_A_Wet_Sunday_Morning.jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-ot-thesis-3.webp" }, "uid": "18qwazf1a6ec1l" }, { "id": "fi-mcq-rw-l2-p1-finchley", "tags": [ "hassabis", "chess", "cognition" ], "title": "What chess taught Hassabis", "body": "Competitive chess gave young Hassabis an early intuition that human cognition has limits — that the bottleneck is hardware, not effort, hinting intelligence might be something you could build.", "shape": "fact", "illustration": { "imageSearchTerm": "young chess prodigy tournament", "imagePrompt": "A young chess player deeply concentrating over a chessboard during a tournament.", "alt": "What chess taught Hassabis", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Young,_Cassin;h92310.jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-mcq-rw-l2-p1-finchley.webp" }, "uid": "1hgmkpj1ecdtjt" }, { "id": "fi-ot-origins-2", "tags": [ "hassabis", "family" ], "title": "His mother's heritage", "body": "Demis Hassabis's mother was Chinese Singaporean. His father was Greek Cypriot, giving Hassabis a mixed heritage often noted in profiles of his early life.", "shape": "fact", "illustration": { "imageSearchTerm": "mixed cultural heritage family", "imagePrompt": "A softly lit image blending two cultural motifs, East Asian and Mediterranean, symbolizing a mixed family heritage.", "alt": "His mother's heritage", "depictable": false, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Mixed_Measures_(33691490850).jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-ot-origins-2.webp" }, "uid": "s380721efeygw" }, { "id": "fi-mcq-theme-park", "tags": [ "hassabis", "biography" ], "title": "Co-designing Theme Park", "body": "As a teenager, Demis Hassabis co-designed the Bullfrog game Theme Park.", "shape": "fact", "illustration": { "imageSearchTerm": "Theme Park video game screenshot", "imagePrompt": "A retro 1990s video game screenshot showing a colorful pixelated amusement park management simulation.", "alt": "Co-designing Theme Park", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Theme_for_embedded_world_2010_(4725843569).jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-mcq-theme-park.webp" }, "uid": "51iso2zto9p0" }, { "id": "fi-ot-games-2", "tags": [ "molyneux", "mentors" ], "title": "Mentor: Peter Molyneux", "body": "Peter Molyneux, already a celebrity in game design, ran the studio that hired the young Hassabis and became one of his most important mentors.", "shape": "fact", "illustration": { "imageSearchTerm": "video game designer conference", "imagePrompt": "A well-known video game designer speaking casually on stage at an industry conference.", "alt": "Mentor: Peter Molyneux", "depictable": true }, "uid": "1on4o8z1y0elyl" }, { "id": "fi-mcq-rw-l3-p1-cambridge", "tags": [ "hassabis", "neuroscience", "deepmind" ], "title": "Brain as generative model", "body": "Hassabis's UCL neuroscience PhD found that patients with hippocampal damage couldn't imagine future scenes, convincing him the brain works as a generative model that builds scenes rather than a lookup table that retrieves them.", "shape": "fact", "illustration": { "imageSearchTerm": "brain MRI scan hippocampus", "imagePrompt": "A detailed brain MRI scan with the hippocampus region highlighted in color.", "alt": "Brain as generative model", "depictable": true, "credit": "Pexels · Close-up of an MRI scan showing a sagittal view of the human brain for analysis.", "creditUrl": "https://www.pexels.com/photo/technology-computer-head-health-7089331/", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-mcq-rw-l3-p1-cambridge.webp" }, "uid": "1s4c4qxzbvobv" }, { "id": "fi-mcq-republic", "tags": [ "elixir" ], "title": "Republic: The Revolution", "body": "Elixir's flagship political simulation was Republic: The Revolution, a 2003 release modelling thousands of citizens with their own goals — critics admired its ambition even though the AI overshot what the era's hardware could handle.", "shape": "fact", "illustration": { "imageSearchTerm": "political strategy video game", "imagePrompt": "A screenshot of an early-2000s political simulation video game showing a stylized city map with citizen icons.", "alt": "Republic: The Revolution", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Manuel_Jorge_-_Pintor_(8).jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-mcq-republic.webp" }, "uid": "q8jagy1ej08h8" }, { "id": "fi-ot-elixir-founding-2", "tags": [ "elixir", "founding" ], "title": "Founded Elixir at 22", "body": "Demis Hassabis was just 22 when he founded Elixir Studios, a year after graduating from Cambridge in 1997.", "shape": "fact", "illustration": { "imageSearchTerm": "1990s video game studio office", "imagePrompt": "A small, scrappy video game studio office in the late 1990s, computers and concept art covering the walls.", "alt": "Founded Elixir at 22", "depictable": true, "credit": "Pexels · Nostalgic collection of vintage video game cartridges showcasing iconic retro gaming titles.", "creditUrl": "https://www.pexels.com/photo/close-up-photo-of-sega-mega-collections-9140592/", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-ot-elixir-founding-2.webp" }, "uid": "tuehmgi9azei" }, { "id": "fi-ot-legg-agi-3", "tags": [ "idsia", "schmidhuber" ], "title": "Legg's IDSIA collaborator", "body": "At the Swiss AI institute IDSIA, Shane Legg worked with Jürgen Schmidhuber on recurrent neural networks, and separately with Marcus Hutter on theories of universal intelligence.", "shape": "fact", "illustration": { "imageSearchTerm": "AI researcher whiteboard equations", "imagePrompt": "A computer scientist at a whiteboard covered in neural network diagrams and equations.", "alt": "Legg's IDSIA collaborator", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Friendship_3.jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-ot-legg-agi-3.webp" }, "uid": "1ly9sm81mgyqkq" }, { "id": "fi-mcq-rw-l4-p1-legg", "tags": [ "legg", "agi", "terminology" ], "title": "A book title brainstorm", "body": "The term 'Artificial General Intelligence' was born around 2002, when Shane Legg and Marcus Hutter were brainstorming a title for a planned collection of essays.", "shape": "fact", "illustration": { "imageSearchTerm": "researchers brainstorming whiteboard", "imagePrompt": "Two researchers at a whiteboard covered with scribbled notes, brainstorming ideas together in an office.", "alt": "A book title brainstorm", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Ahmad_Khalifah.jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-mcq-rw-l4-p1-legg.webp" }, "uid": "lz6bgz1iziml" }, { "id": "fi-ot-suleyman-2", "tags": [ "suleyman", "dates" ], "title": "Suleyman's birth year", "body": "Mustafa Suleyman was born in August 1984 in north London — a detail easy to mix up with cofounder Shane Legg's birth year of 1973.", "shape": "fact", "illustration": { "imageSearchTerm": "Mustafa Suleyman portrait", "imagePrompt": "A professional portrait of a tech entrepreneur and AI executive in business attire.", "alt": "Suleyman's birth year", "depictable": true }, "uid": "1du7y0y26rc4c" }, { "id": "fi-mcq-rw-l5-p1-thiel", "tags": [ "deepmind", "thiel", "funding" ], "title": "DeepMind's first investor", "body": "Peter Thiel's Founders Fund led DeepMind's earliest outside funding round — a notable irony, given Thiel's later public skepticism about unchecked AI development.", "shape": "fact", "illustration": { "imageSearchTerm": "venture capitalist conference portrait", "imagePrompt": "A venture capitalist in a business suit speaking at a technology investment conference.", "alt": "DeepMind's first investor", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:SDCC_2010_Adult_Swim_panel_01.jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-mcq-rw-l5-p1-thiel.webp" }, "uid": "16zol4bmwanpl" }, { "id": "fi-ot-acq-3", "tags": [ "acquisition", "price" ], "title": "Google's $600M deal", "body": "Google's purchase of DeepMind is often quoted as roughly $600 million in US dollars — equivalent to about £400 million.", "shape": "fact", "illustration": { "imageSearchTerm": "tech corporate campus building", "imagePrompt": "A modern corporate campus building with colorful accent architecture and landscaped courtyards.", "alt": "Google's $600M deal", "depictable": true }, "uid": "1l5uxz71qyor1h" }, { "id": "fi-ot-dqn-4", "tags": [ "atari", "reinforcement-learning" ], "title": "Atari's one-network rule", "body": "The Atari challenge imposed a deliberate constraint: a single network, no game-specific code, learning only from raw pixels and the score — forcing a general solution instead of a tailored one.", "shape": "fact", "illustration": { "imageSearchTerm": "retro Atari arcade game", "imagePrompt": "A retro Atari arcade video game screen showing simple blocky graphics, like Breakout or Space Invaders.", "alt": "Atari's one-network rule", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Dawn_Breaks_Theatrical_Poster_-_300dpi_RGB_(1).jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-ot-dqn-4.webp" }, "uid": "d34sintfxsrt" }, { "id": "fi-mcq-alphago-nature", "tags": [ "alphago", "milestones" ], "title": "AlphaGo's Nature cover", "body": "AlphaGo's first major research paper appeared on the cover of the journal Nature in January 2016.", "shape": "fact", "illustration": { "imageSearchTerm": "Go board black white stones", "imagePrompt": "An elegant wooden Go board covered with black and white stones mid-game, viewed from above.", "alt": "AlphaGo's Nature cover", "depictable": true, "credit": "Pexels · Close-up of a Go board game setup with black and white stones and wooden bowls. Perfect for strategy enthusiasts.", "creditUrl": "https://www.pexels.com/photo/strategic-go-board-game-with-stones-and-bowls-38832998/", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-mcq-alphago-nature.webp" }, "uid": "jahj8m9st1ig" }, { "id": "fi-ot-alphago-define-2", "tags": [ "alphago", "architecture" ], "title": "Neural nets plus tree search", "body": "AlphaGo paired deep neural networks with Monte Carlo tree search — a combination that succeeded where brute-force search had failed for decades against Go's enormous search space.", "shape": "fact", "illustration": { "imageSearchTerm": "Go board computer analysis", "imagePrompt": "A wooden Go board with stones, overlaid with faint glowing branching lines suggesting a computer's search process.", "alt": "Neural nets plus tree search", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Go_Nagai_20080704_Japan_Expo_02.jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-ot-alphago-define-2.webp" }, "uid": "14urwprca4dux" }, { "id": "fi-ot-alphago-define-3", "tags": [ "alphago", "lee-sedol", "seoul" ], "title": "Seoul, March 2016: 4-1", "body": "AlphaGo defeated Lee Sedol 4–1 in Seoul in March 2016, a very different result from the 5–0 sweep of Fan Hui in the earlier closed-door match.", "shape": "fact", "illustration": { "imageSearchTerm": "Lee Sedol Go tournament", "imagePrompt": "A professional Go player sitting across a Go board in a formal tournament hall, cameras in the background.", "alt": "Seoul, March 2016: 4-1", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Filming_for_the_movie_Lee_at_K%C3%A1rolyi_Palace_in_Budapest,_October_2022.jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-ot-alphago-define-3.webp" }, "uid": "ltih3k1vfwmn2" }, { "id": "fi-mcq-microsoft-ai-ceo", "tags": [ "suleyman", "race" ], "title": "Suleyman leads Microsoft AI", "body": "Mustafa Suleyman — one of DeepMind's three cofounders — became CEO of Microsoft AI in 2024.", "shape": "fact", "illustration": { "imageSearchTerm": "Microsoft headquarters Redmond", "imagePrompt": "A modern glass corporate campus building exterior, tech industry headquarters aesthetic.", "alt": "Suleyman leads Microsoft AI", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:2009_ticactives_appel_exp%C3%A9rimentations_boustouller_leger_enaud.JPG", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-mcq-microsoft-ai-ceo.webp" }, "uid": "1xz081815mcgze" }, { "id": "fi-ot-three-giants-4", "tags": [ "crisis", "employees", "openai" ], "title": "700 staff threatened to quit", "body": "More than 700 of OpenAI's 770 employees signed a letter threatening to resign within 48 hours of Sam Altman's firing — a near-total revolt that made the board's position untenable.", "shape": "fact", "illustration": { "imageSearchTerm": "tech office employees solidarity", "imagePrompt": "A large group of employees gathered together outside a modern glass office building, standing in solidarity.", "alt": "700 staff threatened to quit", "depictable": true }, "uid": "1ilsm8ycsrfa" }, { "id": "fi-ot-chatgpt-shock-2", "tags": [ "transformer", "dates" ], "title": "Five years before ChatGPT", "body": "The Transformer paper — the architecture ChatGPT ultimately rests on — was published in 2017, five years before ChatGPT's launch put it in front of the general public.", "shape": "fact", "illustration": { "imageSearchTerm": "neural network architecture diagram", "imagePrompt": "An abstract technical diagram of interconnected neural network layers glowing on a dark background.", "alt": "Five years before ChatGPT", "depictable": true, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Neural_Magazine_Cover,_Issue_2_-_Dream_Machine_-_enwiki_version.jpg", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-ot-chatgpt-shock-2.webp" }, "uid": "19d5ffy1o1le9w" }, { "id": "fi-ot-chatgpt-shock-4", "tags": [ "chatgpt", "openai" ], "title": "The model behind ChatGPT", "body": "ChatGPT launched as a polished interface on top of GPT-3.5, presented at the time as a research preview rather than a finished product.", "shape": "fact", "illustration": { "imageSearchTerm": "AI chatbot interface laptop", "imagePrompt": "A laptop screen showing a clean, minimalist chat interface with softly blurred text bubbles.", "alt": "The model behind ChatGPT", "depictable": true, "credit": "Pexels · Close-up of a laptop displaying an AI interface with a chatbot prompt in dark mode.", "creditUrl": "https://www.pexels.com/photo/ai-interface-on-laptop-screen-in-dark-mode-30530412/", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-ot-chatgpt-shock-4.webp" }, "uid": "orosciposj98" }, { "id": "fi-ot-gemini-define-3", "tags": [ "gemini-ultra", "launch" ], "title": "Gemini Ultra's preview-only debut", "body": "At Gemini's launch, two smaller variants shipped immediately, but the headline Gemini Ultra was only previewed — benchmarked against GPT-4 rather than released to the public right away.", "shape": "fact", "illustration": { "imageSearchTerm": "AI model benchmark comparison", "imagePrompt": "An abstract bar chart graphic comparing the glowing heights of two bars, symbolizing a performance benchmark.", "alt": "Gemini Ultra's preview-only debut", "depictable": false }, "uid": "1gv9e6s1c8kbda" }, { "id": "fi-ot-gemini15-3", "tags": [ "mixture-of-experts", "architecture" ], "title": "Only a fraction fires", "body": "Gemini's sparse mixture-of-experts architecture means only a fraction of the model's parameters activate for any given input, which keeps inference costs manageable despite its overall size.", "shape": "fact", "illustration": { "imageSearchTerm": "sparse neural network activation", "imagePrompt": "An abstract glowing network diagram where only a few connected nodes light up while most remain dim.", "alt": "Only a fraction fires", "depictable": false, "credit": "Wikimedia Commons · see source", "creditUrl": "https://commons.wikimedia.org/wiki/File:Ecoregion_PA0603.png", "url": "https://cdn.recurxive.com/packs/infinity-machine/images/fi-ot-gemini15-3.webp" }, "uid": "18h732310vt5ep" }, { "id": "ot-cov-obj-gemini15-build-1", "shape": "mcq", "tags": [ "gemini", "deepmind", "timeline" ], "prompt": { "modality": "text", "value": "In what year did Google DeepMind release Gemini 1.5 Pro?" }, "options": [ { "modality": "text", "value": "2025" }, { "modality": "text", "value": "2023" }, { "modality": "text", "value": "2022" }, { "modality": "text", "value": "2024" } ], "correctIndex": 3, "explanation": "Google DeepMind released Gemini 1.5 Pro on 15 February 2024, about two months after the original Gemini launch.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "o4a4tq1djo9mc" }, { "id": "ot-cov-obj-gemini15-build-2", "shape": "mcq", "tags": [ "gemini", "deepmind", "release-date" ], "prompt": { "modality": "text", "value": "On what date did Google DeepMind release Gemini 1.5 Pro?" }, "options": [ { "modality": "text", "value": "6 December 2023" }, { "modality": "text", "value": "15 February 2024" }, { "modality": "text", "value": "17 November 2023" }, { "modality": "text", "value": "22 January 2024" } ], "correctIndex": 1, "explanation": "Gemini 1.5 Pro shipped on 15 February 2024, built by a London-led strike team of the kind that produced AlphaFold.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "cg1p0ulpya50" }, { "id": "ot-cov-obj-money-rivals-1", "shape": "mcq", "tags": [ "deepmind", "acquisition", "money" ], "prompt": { "modality": "text", "value": "About how much did Google pay to acquire DeepMind in 2014, per Mallaby?" }, "options": [ { "modality": "text", "value": "£800 million" }, { "modality": "text", "value": "£40 million" }, { "modality": "text", "value": "£4 billion" }, { "modality": "text", "value": "£400 million" } ], "correctIndex": 3, "explanation": "Larry Page and Sergey Brin brought DeepMind into Google in 2014 for roughly £400 million.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1kcdsgc5eqyh6" }, { "id": "ot-cov-obj-money-rivals-2", "shape": "mcq", "tags": [ "deepmind", "acquisition", "money" ], "prompt": { "modality": "text", "value": "Which figure is closest to the price tag of Google's 2014 DeepMind deal?" }, "options": [ { "modality": "text", "value": "£4 billion" }, { "modality": "text", "value": "£4 million" }, { "modality": "text", "value": "£400 million" }, { "modality": "text", "value": "£40 million" } ], "correctIndex": 2, "explanation": "The 2014 Google-DeepMind deal, led by Larry Page and Sergey Brin, was worth roughly £400 million.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "pmy1eduvjw0v" }, { "id": "ot-cov-obj-alphago-define-1", "shape": "mcq", "tags": [ "alphago", "nature", "publication" ], "prompt": { "modality": "text", "value": "AlphaGo's 5-0 win over Fan Hui stayed secret until it appeared as a cover story in which publication?" }, "options": [ { "modality": "text", "value": "Science" }, { "modality": "text", "value": "Wired" }, { "modality": "text", "value": "MIT Technology Review" }, { "modality": "text", "value": "Nature" } ], "correctIndex": 3, "explanation": "The Fan Hui result was embargoed until the January 2016 Nature cover paper that announced AlphaGo.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "17te5fe1xju1lg" }, { "id": "ot-cov-obj-alphago-define-2", "shape": "mcq", "tags": [ "alphago", "nature", "timeline" ], "prompt": { "modality": "text", "value": "In what month and year did Nature run the cover paper announcing AlphaGo?" }, "options": [ { "modality": "text", "value": "January 2016" }, { "modality": "text", "value": "October 2015" }, { "modality": "text", "value": "March 2016" }, { "modality": "text", "value": "February 2015" } ], "correctIndex": 0, "explanation": "AlphaGo's Fan Hui result was embargoed until Nature's January 2016 cover paper, which also revealed AlphaGo's architecture.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1817lvxqhpqzv" }, { "id": "ot-cov-obj-openai-founding-1", "shape": "mcq", "tags": [ "openai", "deepmind", "rivalry" ], "prompt": { "modality": "text", "value": "How did Mallaby frame the December 2015 founding of OpenAI?" }, "options": [ { "modality": "text", "value": "As a counterweight to Google's DeepMind" }, { "modality": "text", "value": "As a spinoff of DeepMind's safety team" }, { "modality": "text", "value": "As a merger with DeepMind" }, { "modality": "text", "value": "As a government-run AI lab" } ], "correctIndex": 0, "explanation": "Mallaby frames OpenAI's founding as a non-profit counterweight to Google's ownership of DeepMind.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "o4dfp2f4v8mw" }, { "id": "ot-cov-obj-openai-founding-2", "shape": "mcq", "tags": [ "openai", "musk", "rivalry" ], "prompt": { "modality": "text", "value": "Who concluded that DeepMind 'under Google' was too dangerous a concentration of AI power, driving him to help found OpenAI?" }, "options": [ { "modality": "text", "value": "Sergey Brin" }, { "modality": "text", "value": "Sam Altman" }, { "modality": "text", "value": "Elon Musk" }, { "modality": "text", "value": "Larry Page" } ], "correctIndex": 2, "explanation": "Musk decided a single commercial giant controlling frontier AI was dangerous, with DeepMind under Google his primary concern.", "source": { "label": "The Infinity Machine (book) — Objective Test" }, "uid": "1jkdpx817i6dyu" } ], "lessons": [ { "id": "l1-sweetness", "title": "The Sweetness — Why this book matters", "order": 1, "studyGuidePath": "/packs/infinity-machine/guides/l1-p1-frame.md", "parts": [ { "id": "l1-p1-frame", "title": "Opening the book", "order": 1, "itemIds": [ "fact-sweetness", "fact-infinity-machine", "pair-author", "pair-subject", "mcq-publisher", "tf-f-infinity-machine-pair-author", "tf-t-infinity-machine-pair-subject", "concept-rw-infinity-machine-the-sweetness", "concept-rw-infinity-machine-nobel-prize-in-chemistry", "ot-book-basics-1", "ot-book-basics-2", "ot-book-basics-3", "ot-book-basics-4", "ot-sweetness-1", "ot-sweetness-2", "ot-sweetness-3", "ot-sweetness-4", "ot-sweetness-5", "ot-agi-1", "ot-agi-4", "ot-agi-5", "fi-ot-book-basics-1", "fi-ot-book-basics-2", "fi-mcq-publisher" ], "studyGuideAnchor": "opening-the-book", "studyGuidePath": "/packs/infinity-machine/guides/l1-p1-frame.md" }, { "id": "l1-p2-thesis", "title": "The book's argument", "order": 2, "itemIds": [ "fact-three-strands", "fact-game-ai", "def-agi", "mcq-frame", "fact-thesis-statement", "cz-infinity-machine-fact-thesis-statement", "tf-d-infinity-machine-def-agi", "concept-rw-infinity-machine-the-infinity-machine", "ot-agi-2", "ot-agi-3", "ot-thesis-1", "ot-thesis-2", "ot-thesis-3", "ot-thesis-4", "ot-agidef-1", "fi-mcq-frame", "fi-ot-thesis-3" ], "studyGuideAnchor": "the-book-s-argument", "studyGuidePath": "/packs/infinity-machine/guides/l1-p2-thesis.md" } ], "objectives": [ { "id": "obj-book-basics", "statement": "Identify the author (Sebastian Mallaby), central subject (Demis Hassabis), and publisher of The Infinity Machine", "demonstrationIds": [ "d-book-attribution", "d-book-tf" ] }, { "id": "obj-sweetness", "statement": "Explain 'the sweetness' as Hassabis's name for the instant a hard problem breaks open, and its tie to the AlphaFold Nobel Prize in Chemistry", "demonstrationIds": [ "d-sweetness" ] }, { "id": "obj-agi-define", "statement": "Define AGI (artificial general intelligence) and recognise 'the infinity machine' as Mallaby's label for it", "demonstrationIds": [ "d-agi" ] }, { "id": "obj-thesis", "statement": "Explain Mallaby's thesis that DeepMind is a research lab pursuing AGI as a primary goal, built by people whose biographies and rivalries determine the outcome", "demonstrationIds": [ "d-thesis" ] } ], "demonstrations": [ { "id": "d-book-attribution", "itemIds": [ "pair-author", "pair-subject", "mcq-publisher" ], "label": "Author, subject, publisher" }, { "id": "d-book-tf", "itemIds": [ "tf-f-infinity-machine-pair-author", "tf-t-infinity-machine-pair-subject" ], "label": "Author vs subject T/F" }, { "id": "d-sweetness", "itemIds": [ "concept-rw-infinity-machine-the-sweetness", "concept-rw-infinity-machine-nobel-prize-in-chemistry" ], "label": "The sweetness & the Nobel" }, { "id": "d-agi", "itemIds": [ "def-agi", "tf-d-infinity-machine-def-agi", "concept-rw-infinity-machine-the-infinity-machine" ], "label": "AGI & the infinity machine" }, { "id": "d-thesis", "itemIds": [ "mcq-frame", "cz-infinity-machine-fact-thesis-statement" ], "label": "Mallaby's thesis" } ], "objectiveTests": [ { "id": "test-book-basics", "objectiveId": "obj-book-basics", "title": "Book basics: author, subject, press", "mcqIds": [ "mcq-publisher", "ot-book-basics-1", "ot-book-basics-2", "ot-book-basics-3", "ot-book-basics-4" ] }, { "id": "test-sweetness", "objectiveId": "obj-sweetness", "title": "The sweetness & the Nobel", "mcqIds": [ "ot-sweetness-1", "ot-sweetness-2", "ot-sweetness-3", "ot-sweetness-4", "ot-sweetness-5" ] }, { "id": "test-agi-define", "objectiveId": "obj-agi-define", "title": "AGI and the infinity machine", "mcqIds": [ "ot-agi-1", "ot-agi-2", "ot-agi-4", "ot-agi-5", "ot-agidef-1" ] }, { "id": "test-thesis", "objectiveId": "obj-thesis", "title": "Mallaby's thesis on DeepMind", "mcqIds": [ "mcq-frame", "ot-thesis-1", "ot-thesis-2", "ot-thesis-3", "ot-thesis-4" ] } ] }, { "id": "l2-destiny", "title": "Destiny — Hassabis as a child prodigy", "order": 2, "studyGuidePath": "/packs/infinity-machine/guides/l2-destiny.md", "parts": [ { "id": "l2-p1-finchley", "title": "Finchley and chess", "order": 1, "itemIds": [ "fact-birthplace", "pair-mother", "pair-father", "fact-chess", "numeric-master-age", "fact-zx-spectrum", "mcq-rw-l2-p1-finchley", "num-hassabis-born", "num-kasparov-beaten", "tf-f-infinity-machine-pair-mother", "tf-t-infinity-machine-pair-father", "concept-rw-infinity-machine-zx-spectrum", "ot-origins-1", "ot-origins-2", "ot-origins-3", "ot-origins-4", "ot-origins-5", "ot-chess-1", "ot-chess-2", "ot-chess-3", "ot-chess-4", "ot-formative-1", "ot-formative-3", "ot-formative-5", "fi-mcq-rw-l2-p1-finchley", "fi-ot-origins-2" ], "studyGuideAnchor": "finchley-and-chess" }, { "id": "l2-p2-bullfrog", "title": "Bullfrog and the games path", "order": 2, "itemIds": [ "pair-bullfrog", "mcq-theme-park", "fact-geb", "fact-chess-ceiling", "fact-pdp-1991", "tf-f-infinity-machine-pair-bullfrog", "concept-rw-infinity-machine-theme-park", "concept-rw-infinity-machine-g-del-escher-bach", "concept-rw-infinity-machine-bullfrog-productions", "ot-formative-2", "ot-formative-4", "ot-games-1", "ot-games-2", "ot-games-3", "ot-games-4", "fi-mcq-theme-park", "fi-ot-games-2" ], "studyGuideAnchor": "bullfrog-and-the-games-path" } ], "objectives": [ { "id": "obj-origins", "statement": "Identify Hassabis's 1976 birth and his Chinese Singaporean and Greek Cypriot parentage", "demonstrationIds": [ "d-origins", "d-origins-tf" ] }, { "id": "obj-chess", "statement": "Explain Hassabis's chess prodigy years — master strength by 13, the Deep Blue vs Kasparov landmark, and his intuition that cognition has hardware ceilings", "demonstrationIds": [ "d-chess" ] }, { "id": "obj-formative", "statement": "Recognise the ZX Spectrum and Gödel, Escher, Bach as the formative computer and book behind Hassabis's early belief that intelligence could be built", "demonstrationIds": [ "d-formative" ] }, { "id": "obj-games-path", "statement": "Trace Hassabis's move into game design, co-designing Theme Park at Peter Molyneux's Bullfrog studio", "demonstrationIds": [ "d-bullfrog", "d-studio" ] } ], "demonstrations": [ { "id": "d-origins", "itemIds": [ "pair-mother", "pair-father", "num-hassabis-born" ], "label": "Birth & parentage" }, { "id": "d-origins-tf", "itemIds": [ "tf-f-infinity-machine-pair-mother", "tf-t-infinity-machine-pair-father" ], "label": "Parentage T/F" }, { "id": "d-chess", "itemIds": [ "numeric-master-age", "mcq-rw-l2-p1-finchley", "num-kasparov-beaten" ], "label": "Chess prodigy" }, { "id": "d-formative", "itemIds": [ "concept-rw-infinity-machine-zx-spectrum", "concept-rw-infinity-machine-g-del-escher-bach" ], "label": "Formative influences" }, { "id": "d-bullfrog", "itemIds": [ "pair-bullfrog", "mcq-theme-park", "tf-f-infinity-machine-pair-bullfrog" ], "label": "Bullfrog & Theme Park" }, { "id": "d-studio", "itemIds": [ "concept-rw-infinity-machine-bullfrog-productions", "concept-rw-infinity-machine-theme-park" ], "label": "Studio & its game" } ], "objectiveTests": [ { "id": "test-origins", "objectiveId": "obj-origins", "title": "Birth and parentage", "mcqIds": [ "ot-origins-1", "ot-origins-2", "ot-origins-3", "ot-origins-4", "ot-origins-5" ] }, { "id": "test-chess", "objectiveId": "obj-chess", "title": "The chess prodigy years", "mcqIds": [ "mcq-rw-l2-p1-finchley", "ot-chess-1", "ot-chess-2", "ot-chess-3", "ot-chess-4" ] }, { "id": "test-formative", "objectiveId": "obj-formative", "title": "The computer and the book", "mcqIds": [ "ot-formative-1", "ot-formative-2", "ot-formative-3", "ot-formative-4", "ot-formative-5" ] }, { "id": "test-games-path", "objectiveId": "obj-games-path", "title": "Bullfrog and Theme Park", "mcqIds": [ "mcq-theme-park", "ot-games-1", "ot-games-2", "ot-games-3", "ot-games-4" ] } ] }, { "id": "l3-cambridge-elixir", "title": "Cambridge and Elixir", "order": 3, "studyGuidePath": "/packs/infinity-machine/guides/l3-p1-cambridge.md", "parts": [ { "id": "l3-p1-cambridge", "title": "Cambridge: deep philosophical questions", "order": 1, "itemIds": [ "fact-cambridge", "pair-david-silver", "fact-silver-meeting", "fact-deep-philosophical", "fact-ucl-phd", "mcq-rw-l3-p1-cambridge", "cz-geb-book", "concept-rw-infinity-machine-go", "concept-rw-infinity-machine-hippocampus", "concept-rw-infinity-machine-generative-model", "ot-cambridge-1", "ot-cambridge-2", "ot-cambridge-3", "ot-cambridge-4", "ot-cambridge-5", "ot-neuro-1", "ot-neuro-2", "ot-neuro-3", "ot-neuro-4", "fi-mcq-rw-l3-p1-cambridge" ], "studyGuideAnchor": "cambridge-deep-philosophical-questions", "studyGuidePath": "/packs/infinity-machine/guides/l3-p1-cambridge.md" }, { "id": "l3-p2-elixir", "title": "Elixir Studios", "order": 2, "itemIds": [ "fact-elixir-founded", "numeric-elixir-year", "pair-elixir-game", "mcq-republic", "fact-elixir-end", "fact-elixir-lesson", "def-elixir-studios", "cz-elixir-year", "cz-infinity-machine-def-elixir-studios", "tf-d-infinity-machine-def-elixir-studios", "ot-elixir-founding-1", "ot-elixir-founding-2", "ot-elixir-founding-3", "ot-elixir-founding-4", "ot-elixir-founding-5", "ot-elixir-games-1", "ot-elixir-games-2", "ot-elixir-games-3", "ot-elixir-games-4", "fi-mcq-republic", "fi-ot-elixir-founding-2" ], "studyGuideAnchor": "elixir-studios", "studyGuidePath": "/packs/infinity-machine/guides/l3-p2-elixir.md" } ], "objectives": [ { "id": "obj-cambridge", "statement": "Identify the Cambridge influences on Hassabis — his friendship with David Silver, the board game Go, and the book Gödel, Escher, Bach", "demonstrationIds": [ "d-cambridge" ] }, { "id": "obj-neuroscience", "statement": "Explain Hassabis's UCL neuroscience finding — hippocampal-damage patients couldn't imagine future scenes — leading him to view the brain as a generative model", "demonstrationIds": [ "d-neuro" ] }, { "id": "obj-elixir-founding", "statement": "Recall that Hassabis founded Elixir Studios in 1998 (aged 22) and define what Elixir Studios was", "demonstrationIds": [ "d-elixir-year", "d-elixir-def" ] }, { "id": "obj-elixir-games", "statement": "Identify Elixir's two best-known games — the political simulation Republic: The Revolution and the spy-villain game Evil Genius", "demonstrationIds": [ "d-elixir-games" ] } ], "demonstrations": [ { "id": "d-cambridge", "itemIds": [ "pair-david-silver", "concept-rw-infinity-machine-go", "cz-geb-book" ], "label": "Cambridge influences" }, { "id": "d-neuro", "itemIds": [ "mcq-rw-l3-p1-cambridge", "concept-rw-infinity-machine-hippocampus", "concept-rw-infinity-machine-generative-model" ], "label": "Brain as generative model" }, { "id": "d-elixir-year", "itemIds": [ "numeric-elixir-year", "cz-elixir-year" ], "label": "Elixir founded 1998" }, { "id": "d-elixir-def", "itemIds": [ "def-elixir-studios", "cz-infinity-machine-def-elixir-studios", "tf-d-infinity-machine-def-elixir-studios" ], "label": "What Elixir was" }, { "id": "d-elixir-games", "itemIds": [ "pair-elixir-game", "mcq-republic" ], "label": "Elixir's games" } ], "objectiveTests": [ { "id": "test-cambridge", "objectiveId": "obj-cambridge", "title": "Cambridge influences", "mcqIds": [ "ot-cambridge-1", "ot-cambridge-2", "ot-cambridge-3", "ot-cambridge-4", "ot-cambridge-5" ] }, { "id": "test-neuroscience", "objectiveId": "obj-neuroscience", "title": "UCL neuroscience finding", "mcqIds": [ "mcq-rw-l3-p1-cambridge", "ot-neuro-1", "ot-neuro-2", "ot-neuro-3", "ot-neuro-4" ] }, { "id": "test-elixir-founding", "objectiveId": "obj-elixir-founding", "title": "Founding Elixir Studios", "mcqIds": [ "ot-elixir-founding-1", "ot-elixir-founding-2", "ot-elixir-founding-3", "ot-elixir-founding-4", "ot-elixir-founding-5" ] }, { "id": "test-elixir-games", "objectiveId": "obj-elixir-games", "title": "Elixir's two games", "mcqIds": [ "mcq-republic", "ot-elixir-games-1", "ot-elixir-games-2", "ot-elixir-games-3", "ot-elixir-games-4" ] } ] }, { "id": "l4-gang-of-three", "title": "Gang of Three — Legg and Suleyman", "order": 4, "studyGuidePath": "/packs/infinity-machine/guides/l4-gang-of-three.md", "parts": [ { "id": "l4-p1-legg", "title": "Shane Legg and the word AGI", "order": 1, "itemIds": [ "fact-shane-legg", "pair-legg-term", "pair-idsia", "fact-halloween", "fact-cofounders", "mcq-rw-l4-p1-legg", "tf-t-infinity-machine-pair-idsia", "concept-rw-infinity-machine-marcus-hutter", "concept-rw-infinity-machine-j-rgen-schmidhuber", "ot-legg-agi-1", "ot-legg-agi-2", "ot-legg-agi-3", "ot-legg-agi-4", "ot-idsia-1", "ot-idsia-2", "ot-idsia-3", "ot-idsia-4", "ot-idsia-5", "fi-ot-legg-agi-3", "fi-mcq-rw-l4-p1-legg" ], "studyGuideAnchor": "shane-legg-and-the-word-agi" }, { "id": "l4-p2-suleyman", "title": "Mustafa Suleyman", "order": 2, "itemIds": [ "fact-suleyman-bio", "pair-moose", "mcq-third-cofounder", "fact-vic-poker", "fact-singularity-summit", "fact-thiel-pitch", "num-suleyman-born", "tf-f-infinity-machine-pair-moose", "concept-rw-infinity-machine-george-hassabis", "concept-rw-infinity-machine-luke-nosek", "ot-suleyman-1", "ot-suleyman-2", "ot-suleyman-3", "ot-suleyman-4", "ot-connections-1", "ot-connections-2", "ot-connections-3", "ot-connections-4", "ot-connections-5", "fi-ot-suleyman-2" ], "studyGuideAnchor": "mustafa-suleyman" } ], "objectives": [ { "id": "obj-legg-agi", "statement": "Identify Shane Legg as the coiner of 'AGI' (brainstorming a book title with Marcus Hutter, ~2002) alongside collaborator Jürgen Schmidhuber.", "demonstrationIds": [ "d-legg", "d-legg-collab" ] }, { "id": "obj-idsia", "statement": "Recognise IDSIA as the Swiss AI lab where Legg studied with Hutter and Schmidhuber", "demonstrationIds": [ "d-idsia", "d-legg-collab" ] }, { "id": "obj-suleyman", "statement": "Identify Mustafa Suleyman as DeepMind's third cofounder, his nickname 'Moose', and his 1984 birth", "demonstrationIds": [ "d-suleyman" ] }, { "id": "obj-founding-connections", "statement": "Recognise the personal ties behind the founding — George Hassabis, who introduced Suleyman, and investor Luke Nosek, who heard Thiel's 'Demis's destiny' remark", "demonstrationIds": [ "d-connections" ] } ], "demonstrations": [ { "id": "d-legg", "itemIds": [ "pair-legg-term", "mcq-rw-l4-p1-legg" ], "label": "Legg coins AGI" }, { "id": "d-legg-collab", "itemIds": [ "concept-rw-infinity-machine-marcus-hutter", "concept-rw-infinity-machine-j-rgen-schmidhuber" ], "label": "Legg's IDSIA collaborators" }, { "id": "d-idsia", "itemIds": [ "pair-idsia", "tf-t-infinity-machine-pair-idsia" ], "label": "IDSIA" }, { "id": "d-suleyman", "itemIds": [ "mcq-third-cofounder", "pair-moose", "num-suleyman-born", "tf-f-infinity-machine-pair-moose" ], "label": "Suleyman", "requiredCorrect": 4 }, { "id": "d-connections", "itemIds": [ "concept-rw-infinity-machine-george-hassabis", "concept-rw-infinity-machine-luke-nosek" ], "label": "Founding connections" } ], "objectiveTests": [ { "id": "test-legg-agi", "objectiveId": "obj-legg-agi", "title": "Legg and the word AGI", "mcqIds": [ "mcq-rw-l4-p1-legg", "ot-legg-agi-1", "ot-legg-agi-2", "ot-legg-agi-3", "ot-legg-agi-4" ] }, { "id": "test-idsia", "objectiveId": "obj-idsia", "title": "IDSIA — the Swiss AI lab", "mcqIds": [ "ot-idsia-1", "ot-idsia-2", "ot-idsia-3", "ot-idsia-4", "ot-idsia-5" ] }, { "id": "test-suleyman", "objectiveId": "obj-suleyman", "title": "Mustafa \"Moose\" Suleyman", "mcqIds": [ "mcq-third-cofounder", "ot-suleyman-1", "ot-suleyman-2", "ot-suleyman-3", "ot-suleyman-4" ] }, { "id": "test-founding-connections", "objectiveId": "obj-founding-connections", "title": "Ties behind the founding", "mcqIds": [ "ot-connections-1", "ot-connections-2", "ot-connections-3", "ot-connections-4", "ot-connections-5" ] } ] }, { "id": "l5-founding-deepmind", "title": "Founding DeepMind", "order": 5, "studyGuidePath": "/packs/infinity-machine/guides/l5-p1-thiel.md", "parts": [ { "id": "l5-p1-thiel", "title": "Thiel and the pitch", "order": 1, "itemIds": [ "fact-deepmind-name", "fact-google-acquisition", "numeric-acquisition-price", "fact-musk-page", "fact-ethics-board", "q-im-deepmind-firstname", "q-im-deepmind-registered", "num-google-acquires", "q-im-musk-page-dinner", "mcq-rw-l5-p1-thiel", "num-deepmind-founded", "tf-t-infinity-machine-q-im-deepmind-firstname", "concept-rw-infinity-machine-isaac-asimov", "ot-naming-1", "ot-naming-2", "ot-naming-3", "ot-naming-4", "ot-naming-5", "ot-acq-1", "ot-acq-2", "ot-acq-4", "ot-acq-5", "ot-backers-1", "ot-backers-2", "ot-backers-3", "ot-backers-4", "fi-mcq-rw-l5-p1-thiel" ], "studyGuideAnchor": "thiel-and-the-pitch", "studyGuidePath": "/packs/infinity-machine/guides/l5-p1-thiel.md" }, { "id": "l5-p2-google", "title": "Google takes the wheel", "order": 2, "itemIds": [ "fact-atari-team", "pair-mnih", "def-dqn", "def-q-learning", "fact-experience-replay", "fact-pong-day", "def-deepmind", "mcq-rw-l5-p2-google", "num-google-acquires", "num-google-paid", "cz-deepmind-google", "cz-infinity-machine-def-deepmind", "tf-d-infinity-machine-def-dqn", "tf-df-infinity-machine-def-q-learning", "tf-df-infinity-machine-def-deepmind", "concept-rw-infinity-machine-larry-page", "concept-rw-infinity-machine-elon-musk", "ot-acq-3", "ot-dqn-1", "ot-dqn-2", "ot-dqn-3", "ot-dqn-4", "ot-dmdef-1", "ot-dmdef-2", "ot-dmdef-3", "ot-dmdef-4", "ot-dmdef-5", "fi-ot-acq-3", "fi-ot-dqn-4" ], "studyGuideAnchor": "google-takes-the-wheel", "studyGuidePath": "/packs/infinity-machine/guides/l5-p2-google.md" } ], "objectives": [ { "id": "obj-naming-founding", "statement": "Explain the DeepMind name's origin (first named Solaria, after an Asimov novel) and that the lab was founded and registered in 2010", "demonstrationIds": [ "d-naming", "d-founding-year" ] }, { "id": "obj-google-acq", "statement": "Recall Google's 2014 acquisition of DeepMind for roughly £400 million / $600 million and Larry Page's role", "demonstrationIds": [ "d-acq-price", "d-acq-year" ] }, { "id": "obj-early-backers", "statement": "Recognise the Silicon Valley backdrop — Peter Thiel's Founders Fund as first backer and the October 2012 Musk–Page dinner where Musk warned of a 'digital god'", "demonstrationIds": [ "d-thiel-fund", "d-musk-page" ] }, { "id": "obj-atari-dqn", "statement": "Define Deep Q-Networks and Q-learning, identify Vlad Mnih as the Atari DQN lead, and explain why experience replay made deep RL stable", "demonstrationIds": [ "d-dqn-def", "d-dqn-team" ] }, { "id": "obj-deepmind-define", "statement": "Define DeepMind as the London AI lab Hassabis cofounded in 2010 and Google acquired in 2014", "demonstrationIds": [ "d-deepmind-def" ] } ], "demonstrations": [ { "id": "d-naming", "itemIds": [ "q-im-deepmind-firstname", "tf-t-infinity-machine-q-im-deepmind-firstname", "concept-rw-infinity-machine-isaac-asimov" ], "label": "Solaria / Asimov" }, { "id": "d-founding-year", "itemIds": [ "q-im-deepmind-registered", "num-deepmind-founded" ], "label": "Founded 2010" }, { "id": "d-acq-price", "itemIds": [ "numeric-acquisition-price", "num-google-paid", "cz-deepmind-google" ], "label": "Acquisition price" }, { "id": "d-acq-year", "itemIds": [ "num-google-acquires", "concept-rw-infinity-machine-larry-page" ], "label": "Acquisition year & Page" }, { "id": "d-thiel-fund", "itemIds": [ "mcq-rw-l5-p1-thiel", "q-im-musk-page-dinner" ], "label": "Thiel & the dinner" }, { "id": "d-musk-page", "itemIds": [ "q-im-musk-page-dinner", "concept-rw-infinity-machine-elon-musk" ], "label": "Musk's digital god warning" }, { "id": "d-dqn-def", "itemIds": [ "def-dqn", "def-q-learning", "tf-d-infinity-machine-def-dqn", "tf-df-infinity-machine-def-q-learning" ], "label": "DQN & Q-learning", "requiredCorrect": 3 }, { "id": "d-dqn-team", "itemIds": [ "pair-mnih", "mcq-rw-l5-p2-google" ], "label": "Mnih & experience replay" }, { "id": "d-deepmind-def", "itemIds": [ "def-deepmind", "cz-infinity-machine-def-deepmind", "tf-df-infinity-machine-def-deepmind" ], "label": "Define DeepMind" } ], "objectiveTests": [ { "id": "test-naming-founding", "objectiveId": "obj-naming-founding", "title": "The DeepMind name & 2010 founding", "mcqIds": [ "ot-naming-1", "ot-naming-2", "ot-naming-3", "ot-naming-4", "ot-naming-5" ] }, { "id": "test-google-acq", "objectiveId": "obj-google-acq", "title": "Google's 2014 acquisition", "mcqIds": [ "ot-acq-1", "ot-acq-2", "ot-acq-3", "ot-acq-4", "ot-acq-5" ] }, { "id": "test-early-backers", "objectiveId": "obj-early-backers", "title": "Thiel, Musk and Page", "mcqIds": [ "mcq-rw-l5-p1-thiel", "ot-backers-1", "ot-backers-2", "ot-backers-3", "ot-backers-4" ] }, { "id": "test-atari-dqn", "objectiveId": "obj-atari-dqn", "title": "Atari, DQN and experience replay", "mcqIds": [ "mcq-rw-l5-p2-google", "ot-dqn-1", "ot-dqn-2", "ot-dqn-3", "ot-dqn-4" ] }, { "id": "test-deepmind-define", "objectiveId": "obj-deepmind-define", "title": "What DeepMind is", "mcqIds": [ "ot-dmdef-1", "ot-dmdef-2", "ot-dmdef-3", "ot-dmdef-4", "ot-dmdef-5" ] } ] }, { "id": "l6-atari-alphago", "title": "Atari to AlphaGo", "order": 6, "studyGuidePath": "/packs/infinity-machine/guides/l6-p1-atari.md", "parts": [ { "id": "l6-p1-atari", "title": "Atari and the DQN", "order": 1, "itemIds": [ "fact-49-games", "fact-brin-go", "pair-aja", "fact-fan-hui", "fact-lee-sedol", "fact-move-37", "def-reinforcement-learning", "mcq-rw-l6-p1-atari", "num-alexnet", "num-dqn-nature", "num-dqn-games", "num-dqn-human-level", "cz-infinity-machine-def-reinforcement-learning", "tf-t-infinity-machine-pair-aja", "tf-df-infinity-machine-def-reinforcement-learning", "concept-rw-infinity-machine-fan-hui", "concept-rw-infinity-machine-lee-sedol", "concept-rw-infinity-machine-aja-huang", "concept-rw-infinity-machine-sergey-brin", "ot-rl-define-1", "ot-rl-define-2", "ot-rl-define-3", "ot-rl-define-4", "ot-rl-define-5", "ot-go-people-1", "ot-go-people-2", "ot-go-people-3", "ot-go-people-4", "ot-go-people-5", "ot-move37-1", "ot-move37-2", "ot-move37-3", "ot-move37-4", "ot-cov-obj-alphago-define-1" ], "studyGuideAnchor": "atari-and-the-dqn", "studyGuidePath": "/packs/infinity-machine/guides/l6-p1-atari.md" }, { "id": "l6-p2-go", "title": "AlphaGo and Move 37", "order": 2, "itemIds": [ "numeric-alphago-seoul", "mcq-alphago-nature", "fact-openai-founding", "pair-altman", "pair-sutskever", "fact-poaching", "def-alphago", "num-alphago-seoul", "cz-alphago-sedol", "cz-infinity-machine-def-alphago", "tf-f-infinity-machine-pair-altman", "tf-t-infinity-machine-pair-sutskever", "tf-df-infinity-machine-def-alphago", "ot-alphago-define-1", "ot-alphago-define-2", "ot-alphago-define-3", "ot-alphago-define-4", "ot-openai-founding-1", "ot-openai-founding-2", "ot-openai-founding-3", "ot-openai-founding-4", "ot-openai-founding-5", "fi-mcq-alphago-nature", "fi-ot-alphago-define-2", "fi-ot-alphago-define-3", "ot-cov-obj-alphago-define-2", "ot-cov-obj-openai-founding-1", "ot-cov-obj-openai-founding-2" ], "studyGuideAnchor": "alphago-and-move-37", "studyGuidePath": "/packs/infinity-machine/guides/l6-p2-go.md" } ], "objectives": [ { "id": "obj-rl-define", "statement": "Define reinforcement learning and recall the DQN milestones — published in Nature (2015), human-level or better on 29 of 49 Atari games.", "demonstrationIds": [ "d-rl-def", "d-dqn-facts" ] }, { "id": "obj-alphago-define", "statement": "Define AlphaGo as the DeepMind program that beat world champion Lee Sedol at Go in 2016 (announced in Nature)", "demonstrationIds": [ "d-alphago-def", "d-alphago-seoul" ] }, { "id": "obj-go-people", "statement": "Identify the people around AlphaGo — engineer Aja Huang, opponents Fan Hui and Lee Sedol, and Sergey Brin who backed the project", "demonstrationIds": [ "d-go-people", "d-go-opponents" ] }, { "id": "obj-move37", "statement": "Explain Move 37 — the shoulder-hit AlphaGo's policy network rated a roughly 1-in-10,000 human move against Lee Sedol", "demonstrationIds": [ "d-move37" ] }, { "id": "obj-openai-founding", "statement": "Identify OpenAI's founders — CEO Sam Altman and first chief scientist Ilya Sutskever — formed as a counterweight to Google's DeepMind", "demonstrationIds": [ "d-openai" ] } ], "demonstrations": [ { "id": "d-rl-def", "itemIds": [ "def-reinforcement-learning", "cz-infinity-machine-def-reinforcement-learning", "tf-df-infinity-machine-def-reinforcement-learning" ], "label": "Define RL" }, { "id": "d-dqn-facts", "itemIds": [ "num-dqn-nature", "num-dqn-games", "num-dqn-human-level", "num-alexnet" ], "label": "DQN milestones" }, { "id": "d-alphago-def", "itemIds": [ "def-alphago", "cz-infinity-machine-def-alphago", "tf-df-infinity-machine-def-alphago", "ot-cov-obj-alphago-define-1", "ot-cov-obj-alphago-define-2" ], "label": "Define AlphaGo" }, { "id": "d-alphago-seoul", "itemIds": [ "numeric-alphago-seoul", "num-alphago-seoul", "cz-alphago-sedol", "mcq-alphago-nature" ], "label": "Seoul 2016 & Nature", "requiredCorrect": 3 }, { "id": "d-go-people", "itemIds": [ "pair-aja", "tf-t-infinity-machine-pair-aja", "concept-rw-infinity-machine-aja-huang" ], "label": "Aja Huang" }, { "id": "d-go-opponents", "itemIds": [ "concept-rw-infinity-machine-fan-hui", "concept-rw-infinity-machine-lee-sedol", "concept-rw-infinity-machine-sergey-brin" ], "label": "Opponents & backer" }, { "id": "d-move37", "itemIds": [ "mcq-rw-l6-p1-atari", "concept-rw-infinity-machine-lee-sedol" ], "label": "Move 37" }, { "id": "d-openai", "itemIds": [ "pair-altman", "pair-sutskever", "tf-f-infinity-machine-pair-altman", "tf-t-infinity-machine-pair-sutskever", "ot-cov-obj-openai-founding-1", "ot-cov-obj-openai-founding-2" ], "label": "OpenAI founders", "requiredCorrect": 3 } ], "objectiveTests": [ { "id": "test-rl-define", "objectiveId": "obj-rl-define", "title": "Reinforcement Learning & DQN", "mcqIds": [ "ot-rl-define-1", "ot-rl-define-2", "ot-rl-define-3", "ot-rl-define-4", "ot-rl-define-5" ] }, { "id": "test-alphago-define", "objectiveId": "obj-alphago-define", "title": "AlphaGo: What It Was", "mcqIds": [ "mcq-alphago-nature", "ot-alphago-define-1", "ot-alphago-define-2", "ot-alphago-define-3", "ot-alphago-define-4", "ot-cov-obj-alphago-define-1", "ot-cov-obj-alphago-define-2" ] }, { "id": "test-go-people", "objectiveId": "obj-go-people", "title": "The People Around AlphaGo", "mcqIds": [ "ot-go-people-1", "ot-go-people-2", "ot-go-people-3", "ot-go-people-4", "ot-go-people-5" ] }, { "id": "test-move37", "objectiveId": "obj-move37", "title": "Move 37", "mcqIds": [ "mcq-rw-l6-p1-atari", "ot-move37-1", "ot-move37-2", "ot-move37-3", "ot-move37-4" ] }, { "id": "test-openai-founding", "objectiveId": "obj-openai-founding", "title": "The Founding of OpenAI", "mcqIds": [ "ot-openai-founding-1", "ot-openai-founding-2", "ot-openai-founding-3", "ot-openai-founding-4", "ot-openai-founding-5", "ot-cov-obj-openai-founding-1", "ot-cov-obj-openai-founding-2" ] } ] }, { "id": "l7-openai", "title": "The OpenAI rivalry", "order": 7, "studyGuidePath": "/packs/infinity-machine/guides/l7-p1-musk.md", "parts": [ { "id": "l7-p1-musk", "title": "Musk, Altman and the split", "order": 1, "itemIds": [ "fact-musk-email", "fact-restructure-2019", "fact-suleyman-departure", "fact-google-merger", "fact-three-northlondoners", "q-im-brockman-email", "q-im-openai-cap", "q-im-openai-restructure-year", "q-im-google-deepmind-merger", "mcq-rw-l7-p1-musk", "tf-f-infinity-machine-q-im-openai-cap", "tf-f-infinity-machine-q-im-google-deepmind-merger", "concept-rw-infinity-machine-greg-brockman", "ot-openai-split-1", "ot-openai-split-2", "ot-openai-split-3", "ot-openai-split-4", "ot-capped-profit-1", "ot-capped-profit-2", "ot-capped-profit-3", "ot-capped-profit-4", "ot-capped-profit-5", "ot-gdm-merger-1", "ot-gdm-merger-2", "ot-gdm-merger-3", "ot-gdm-merger-4", "ot-gdm-merger-5", "ot-three-giants-1" ], "studyGuideAnchor": "musk-altman-and-the-split", "studyGuidePath": "/packs/infinity-machine/guides/l7-p1-musk.md" }, { "id": "l7-p2-anthropic", "title": "Anthropic and the firing", "order": 2, "itemIds": [ "fact-anthropic", "mcq-microsoft-ai-ceo", "fact-altman-firing", "numeric-altman-days", "fact-nadella-offer", "concept-rw-infinity-machine-dario-amodei", "concept-rw-infinity-machine-satya-nadella", "ot-three-giants-2", "ot-three-giants-3", "ot-three-giants-4", "fi-mcq-microsoft-ai-ceo", "fi-ot-three-giants-4" ], "studyGuideAnchor": "anthropic-and-the-firing", "studyGuidePath": "/packs/infinity-machine/guides/l7-p2-anthropic.md" } ], "objectives": [ { "id": "obj-openai-split", "statement": "Explain the Musk–Altman split — the 2017 email warning of 'absolute control' and Musk's withdrawal after losing operational control.", "demonstrationIds": [ "d-email" ] }, { "id": "obj-capped-profit", "statement": "Recall OpenAI's 2019 restructuring into a capped-profit company with investor returns capped at 100×", "demonstrationIds": [ "d-cap" ] }, { "id": "obj-google-deepmind-merger", "statement": "Identify the April 2023 merger of Google Brain into DeepMind under Hassabis to form Google DeepMind", "demonstrationIds": [ "d-merger" ] }, { "id": "obj-three-giants", "statement": "Recognize how the founders ended up atop rival giants (Suleyman as Microsoft AI CEO) and the players in OpenAI's 5-day crisis.", "demonstrationIds": [ "d-suleyman-ms", "d-crisis-people" ] } ], "demonstrations": [ { "id": "d-email", "itemIds": [ "q-im-brockman-email", "concept-rw-infinity-machine-greg-brockman", "mcq-rw-l7-p1-musk" ], "label": "2017 email & Musk exit", "requiredCorrect": 3 }, { "id": "d-cap", "itemIds": [ "q-im-openai-cap", "q-im-openai-restructure-year", "tf-f-infinity-machine-q-im-openai-cap" ], "label": "Capped-profit 2019" }, { "id": "d-merger", "itemIds": [ "q-im-google-deepmind-merger", "tf-f-infinity-machine-q-im-google-deepmind-merger" ], "label": "Google DeepMind merger" }, { "id": "d-suleyman-ms", "itemIds": [ "mcq-microsoft-ai-ceo", "numeric-altman-days" ], "label": "Suleyman & the 5-day firing" }, { "id": "d-crisis-people", "itemIds": [ "concept-rw-infinity-machine-dario-amodei", "concept-rw-infinity-machine-satya-nadella" ], "label": "Amodei & Nadella" } ], "objectiveTests": [ { "id": "test-openai-split", "objectiveId": "obj-openai-split", "title": "The Musk-Altman split", "mcqIds": [ "mcq-rw-l7-p1-musk", "ot-openai-split-1", "ot-openai-split-2", "ot-openai-split-3", "ot-openai-split-4" ] }, { "id": "test-capped-profit", "objectiveId": "obj-capped-profit", "title": "OpenAI's capped-profit pivot", "mcqIds": [ "ot-capped-profit-1", "ot-capped-profit-2", "ot-capped-profit-3", "ot-capped-profit-4", "ot-capped-profit-5" ] }, { "id": "test-google-deepmind-merger", "objectiveId": "obj-google-deepmind-merger", "title": "The Google DeepMind merger", "mcqIds": [ "ot-gdm-merger-1", "ot-gdm-merger-2", "ot-gdm-merger-3", "ot-gdm-merger-4", "ot-gdm-merger-5" ] }, { "id": "test-three-giants", "objectiveId": "obj-three-giants", "title": "Three giants, five days", "mcqIds": [ "mcq-microsoft-ai-ceo", "ot-three-giants-1", "ot-three-giants-2", "ot-three-giants-3", "ot-three-giants-4" ] } ] }, { "id": "l8-alphafold", "title": "AlphaFold — Fermat for biology", "order": 8, "studyGuidePath": "/packs/infinity-machine/guides/l8-p1-promise.md", "parts": [ { "id": "l8-p1-promise", "title": "A promise after Seoul", "order": 1, "itemIds": [ "fact-alphafold-seoul", "def-protein-folding", "fact-anfinsen", "pair-jumper", "fact-jumper-arrives", "fact-casp", "def-gdt", "fact-distogram", "def-alphafold", "mcq-rw-l8-p1-promise", "cz-infinity-machine-def-alphafold", "tf-d-infinity-machine-def-protein-folding", "tf-d-infinity-machine-def-gdt", "tf-d-infinity-machine-def-alphafold", "concept-rw-infinity-machine-anfinsen-s-conjecture", "concept-rw-infinity-machine-distogram", "ot-folding-1", "ot-folding-2", "ot-folding-3", "ot-folding-4", "ot-folding-5", "ot-afdef-1", "ot-afdef-2", "ot-afdef-3", "ot-afdef-4", "ot-afdef-5", "ot-disto-1", "ot-disto-2", "ot-disto-3", "ot-disto-4" ], "studyGuideAnchor": "a-promise-after-seoul", "studyGuidePath": "/packs/infinity-machine/guides/l8-p1-promise.md" }, { "id": "l8-p2-casp14", "title": "CASP14 and the 200 million", "order": 2, "itemIds": [ "fact-cancun-2018", "fact-direct-folding", "fact-tetraformer", "fact-casp14", "numeric-casp14-score", "fact-database", "numeric-proteins", "fact-cheese-tomorrow", "def-casp", "mcq-rw-l8-p2-casp14", "cz-alphafold-casp", "cz-infinity-machine-def-casp", "tf-d-infinity-machine-def-casp", "concept-rw-infinity-machine-casp", "concept-rw-infinity-machine-gdt-global-distance-test", "ot-gdtcasp-1", "ot-gdtcasp-2", "ot-gdtcasp-3", "ot-gdtcasp-4", "ot-gdtcasp-5", "ot-casp14-1", "ot-casp14-2", "ot-casp14-3", "ot-casp14-4" ], "studyGuideAnchor": "casp14-and-the-200-million", "studyGuidePath": "/packs/infinity-machine/guides/l8-p2-casp14.md" } ], "objectives": [ { "id": "obj-protein-folding", "statement": "Define protein folding and recognise Anfinsen's conjecture that a protein's sequence alone determines its folded shape", "demonstrationIds": [ "d-folding" ] }, { "id": "obj-alphafold-define", "statement": "Define AlphaFold as the DeepMind system that predicts 3D protein structure from amino-acid sequence, and identify John Jumper as AlphaFold 2's lead", "demonstrationIds": [ "d-alphafold-def" ] }, { "id": "obj-distogram", "statement": "Explain AlphaFold 1's leap — predicting the continuous distance between amino-acid pairs (a 'distogram') rather than binary contact", "demonstrationIds": [ "d-distogram" ] }, { "id": "obj-gdt-casp", "statement": "Define the GDT accuracy metric and the biennial CASP contest that scores protein-structure predictions", "demonstrationIds": [ "d-gdt", "d-casp" ] }, { "id": "obj-casp14-results", "statement": "Recall AlphaFold 2's landmark results — 92 GDT at CASP14 (2020), the direct-folding pivot, and ~200 million proteins released by July 2022.", "demonstrationIds": [ "d-casp14" ] } ], "demonstrations": [ { "id": "d-folding", "itemIds": [ "def-protein-folding", "tf-d-infinity-machine-def-protein-folding", "concept-rw-infinity-machine-anfinsen-s-conjecture" ], "label": "Protein folding & Anfinsen" }, { "id": "d-alphafold-def", "itemIds": [ "def-alphafold", "cz-infinity-machine-def-alphafold", "tf-d-infinity-machine-def-alphafold", "pair-jumper" ], "label": "AlphaFold & Jumper", "requiredCorrect": 4 }, { "id": "d-distogram", "itemIds": [ "mcq-rw-l8-p1-promise", "concept-rw-infinity-machine-distogram" ], "label": "Distogram" }, { "id": "d-gdt", "itemIds": [ "def-gdt", "tf-d-infinity-machine-def-gdt", "concept-rw-infinity-machine-gdt-global-distance-test" ], "label": "GDT metric" }, { "id": "d-casp", "itemIds": [ "def-casp", "cz-infinity-machine-def-casp", "tf-d-infinity-machine-def-casp", "concept-rw-infinity-machine-casp" ], "label": "CASP contest", "requiredCorrect": 3 }, { "id": "d-casp14", "itemIds": [ "numeric-casp14-score", "cz-alphafold-casp", "mcq-rw-l8-p2-casp14", "numeric-proteins" ], "label": "CASP14 & 200M", "requiredCorrect": 4 } ], "objectiveTests": [ { "id": "test-protein-folding", "objectiveId": "obj-protein-folding", "title": "Protein folding & Anfinsen", "mcqIds": [ "ot-folding-1", "ot-folding-2", "ot-folding-3", "ot-folding-4", "ot-folding-5" ] }, { "id": "test-alphafold-define", "objectiveId": "obj-alphafold-define", "title": "What AlphaFold is", "mcqIds": [ "ot-afdef-1", "ot-afdef-2", "ot-afdef-3", "ot-afdef-4", "ot-afdef-5" ] }, { "id": "test-distogram", "objectiveId": "obj-distogram", "title": "Contact map to distogram", "mcqIds": [ "mcq-rw-l8-p1-promise", "ot-disto-1", "ot-disto-2", "ot-disto-3", "ot-disto-4" ] }, { "id": "test-gdt-casp", "objectiveId": "obj-gdt-casp", "title": "GDT and CASP", "mcqIds": [ "ot-gdtcasp-1", "ot-gdtcasp-2", "ot-gdtcasp-3", "ot-gdtcasp-4", "ot-gdtcasp-5" ] }, { "id": "test-casp14-results", "objectiveId": "obj-casp14-results", "title": "CASP14 and the 200 million", "mcqIds": [ "mcq-rw-l8-p2-casp14", "ot-casp14-1", "ot-casp14-2", "ot-casp14-3", "ot-casp14-4" ] } ] }, { "id": "l9-racegpt", "title": "RaceGPT and Bletchley", "order": 9, "studyGuidePath": "/packs/infinity-machine/guides/l9-p1-shock.md", "parts": [ { "id": "l9-p1-shock", "title": "The ChatGPT shock", "order": 1, "itemIds": [ "fact-chatgpt-shock", "pair-irving", "fact-irving-arrives", "def-alignment", "fact-one-sentence", "mcq-rw-l9-p1-shock", "num-transformer-paper", "num-chatgpt-launch", "tf-d-infinity-machine-def-alignment", "concept-rw-infinity-machine-yoshua-bengio", "concept-rw-infinity-machine-yann-lecun", "ot-chatgpt-shock-1", "ot-chatgpt-shock-2", "ot-chatgpt-shock-3", "ot-chatgpt-shock-4", "ot-alignment-figures-1", "ot-alignment-figures-2", "ot-alignment-figures-3", "ot-alignment-figures-4", "ot-alignment-figures-5", "fi-ot-chatgpt-shock-2", "fi-ot-chatgpt-shock-4" ], "studyGuideAnchor": "the-chatgpt-shock", "studyGuidePath": "/packs/infinity-machine/guides/l9-p1-shock.md" }, { "id": "l9-p2-governments", "title": "Governments arrive", "order": 2, "itemIds": [ "fact-buchanan", "fact-voluntary-commitments", "fact-executive-order", "fact-bletchley", "numeric-bletchley-countries", "fact-china-policy", "fact-sputnik", "mcq-ai-safety-institute", "q-im-buchanan-czar", "q-im-voluntary-commitments-date", "q-im-bletchley-year", "q-im-chip-ban-year", "cz-bletchley-declaration", "tf-f-infinity-machine-q-im-buchanan-czar", "concept-rw-infinity-machine-ke-jie", "ot-govt-response-1", "ot-govt-response-2", "ot-govt-response-3", "ot-govt-response-4", "ot-govt-response-5", "ot-bletchley-1", "ot-bletchley-2", "ot-bletchley-3", "ot-bletchley-4", "ot-chip-china-1", "ot-chip-china-2", "ot-chip-china-3", "ot-chip-china-4", "ot-chip-china-5" ], "studyGuideAnchor": "governments-arrive", "studyGuidePath": "/packs/infinity-machine/guides/l9-p2-governments.md" } ], "objectives": [ { "id": "obj-chatgpt-shock", "statement": "Recall ChatGPT's 2022 launch and record-fast adoption (100 million users within two months), plus the 2017 Transformer paper behind it", "demonstrationIds": [ "d-chatgpt" ] }, { "id": "obj-alignment-figures", "statement": "Define AI alignment, identify Geoffrey Irving, and recognize safety-letter signatory Yoshua Bengio and non-signatory Yann LeCun.", "demonstrationIds": [ "d-alignment", "d-safety-people" ] }, { "id": "obj-govt-response", "statement": "Trace the US government response — AI czar Ben Buchanan and the July 2023 White House voluntary commitments", "demonstrationIds": [ "d-buchanan", "d-commitments" ] }, { "id": "obj-bletchley", "statement": "Recall the November 2023 Bletchley Park summit — 28 countries signed the Bletchley Declaration, with the UK and US announcing AI Safety Institutes", "demonstrationIds": [ "d-bletchley" ] }, { "id": "obj-chip-china", "statement": "Recognise the 2022 US chip-export ban on China and Ke Jie, the champion whose AlphaGo defeat became China's 'Sputnik moment'", "demonstrationIds": [ "d-china" ] } ], "demonstrations": [ { "id": "d-chatgpt", "itemIds": [ "mcq-rw-l9-p1-shock", "num-chatgpt-launch", "num-transformer-paper" ], "label": "ChatGPT shock" }, { "id": "d-alignment", "itemIds": [ "def-alignment", "tf-d-infinity-machine-def-alignment" ], "label": "Define alignment" }, { "id": "d-safety-people", "itemIds": [ "pair-irving", "concept-rw-infinity-machine-yoshua-bengio", "concept-rw-infinity-machine-yann-lecun" ], "label": "Safety figures" }, { "id": "d-buchanan", "itemIds": [ "q-im-buchanan-czar", "tf-f-infinity-machine-q-im-buchanan-czar" ], "label": "AI czar Buchanan" }, { "id": "d-commitments", "itemIds": [ "q-im-voluntary-commitments-date" ], "label": "White House commitments" }, { "id": "d-bletchley", "itemIds": [ "numeric-bletchley-countries", "mcq-ai-safety-institute", "q-im-bletchley-year", "cz-bletchley-declaration" ], "label": "Bletchley summit", "requiredCorrect": 4 }, { "id": "d-china", "itemIds": [ "q-im-chip-ban-year", "concept-rw-infinity-machine-ke-jie" ], "label": "Chip ban & Ke Jie" } ], "objectiveTests": [ { "id": "test-chatgpt-shock", "objectiveId": "obj-chatgpt-shock", "title": "The ChatGPT shock", "mcqIds": [ "mcq-rw-l9-p1-shock", "ot-chatgpt-shock-1", "ot-chatgpt-shock-2", "ot-chatgpt-shock-3", "ot-chatgpt-shock-4" ] }, { "id": "test-alignment-figures", "objectiveId": "obj-alignment-figures", "title": "Alignment and the safety letter", "mcqIds": [ "ot-alignment-figures-1", "ot-alignment-figures-2", "ot-alignment-figures-3", "ot-alignment-figures-4", "ot-alignment-figures-5" ] }, { "id": "test-govt-response", "objectiveId": "obj-govt-response", "title": "The US government response", "mcqIds": [ "ot-govt-response-1", "ot-govt-response-2", "ot-govt-response-3", "ot-govt-response-4", "ot-govt-response-5" ] }, { "id": "test-bletchley", "objectiveId": "obj-bletchley", "title": "Bletchley Park summit", "mcqIds": [ "mcq-ai-safety-institute", "ot-bletchley-1", "ot-bletchley-2", "ot-bletchley-3", "ot-bletchley-4" ] }, { "id": "test-chip-china", "objectiveId": "obj-chip-china", "title": "Chips, China and Ke Jie", "mcqIds": [ "ot-chip-china-1", "ot-chip-china-2", "ot-chip-china-3", "ot-chip-china-4", "ot-chip-china-5" ] } ] }, { "id": "l10-comeback", "title": "Step by Step — Gemini and the comeback", "order": 10, "studyGuidePath": "/packs/infinity-machine/guides/l10-p1-gemini.md", "parts": [ { "id": "l10-p1-gemini", "title": "Gemini ships", "order": 1, "itemIds": [ "fact-gemini-launch", "def-mmlu", "numeric-gemini-mmlu", "fact-merger-culture", "fact-silver-quits-rlhf", "fact-mech-interp", "def-gemini", "mcq-rw-l10-p1-gemini", "cz-infinity-machine-def-gemini", "tf-d-infinity-machine-def-mmlu", "tf-d-infinity-machine-def-gemini", "concept-rw-infinity-machine-gemini-ultra", "concept-rw-infinity-machine-mmlu", "ot-gemini-define-1", "ot-gemini-define-2", "ot-gemini-define-3", "ot-gemini-define-4", "ot-gemini-define-5", "ot-mmlu-1", "ot-mmlu-2", "ot-mmlu-3", "ot-mmlu-4", "ot-mmlu-5", "ot-gemini15-4", "fi-ot-gemini-define-3" ], "studyGuideAnchor": "gemini-ships", "studyGuidePath": "/packs/infinity-machine/guides/l10-p1-gemini.md" }, { "id": "l10-p2-end", "title": "Mallaby's closing register", "order": 2, "itemIds": [ "fact-gemini-15", "fact-still-first-innings", "fact-altman-untrustworthy", "fact-conclusion", "q-im-gemini15-release", "q-im-gemini15-year", "q-im-altman-firing-triggers", "q-im-hassabis-nobel", "mcq-rw-l10-p2-end", "tf-f-infinity-machine-q-im-gemini15-release", "tf-f-infinity-machine-q-im-altman-firing-triggers", "tf-f-infinity-machine-q-im-hassabis-nobel", "concept-rw-infinity-machine-gemini-1-5-pro", "concept-rw-infinity-machine-sparse-mixture-of-experts", "ot-gemini15-1", "ot-gemini15-2", "ot-gemini15-3", "ot-firing-1", "ot-firing-2", "ot-firing-3", "ot-firing-4", "ot-firing-5", "ot-closing-1", "ot-closing-2", "ot-closing-3", "ot-closing-4", "fi-ot-gemini15-3", "ot-cov-obj-gemini15-build-1", "ot-cov-obj-gemini15-build-2" ], "studyGuideAnchor": "mallaby-s-closing-register", "studyGuidePath": "/packs/infinity-machine/guides/l10-p2-end.md" } ], "objectives": [ { "id": "obj-gemini-define", "statement": "Define Gemini as Google DeepMind's flagship multimodal model family (launched 2023) and identify Gemini Ultra", "demonstrationIds": [ "d-gemini-def" ] }, { "id": "obj-mmlu-score", "statement": "Explain the MMLU benchmark (57 subjects) and Gemini Ultra's announced 90% score", "demonstrationIds": [ "d-mmlu" ] }, { "id": "obj-gemini15-build", "statement": "Recall Gemini 1.5 Pro (2024) with its million-token context window and sparse mixture-of-experts design, and why RL couldn't crack post-training", "demonstrationIds": [ "d-gemini15", "d-build" ] }, { "id": "obj-altman-firing-triggers", "statement": "Identify the two triggers Mallaby gives for the November 2023 Altman firing — undisclosed Gulf fundraising and pushing out critical board members", "demonstrationIds": [ "d-firing" ] }, { "id": "obj-closing", "statement": "Explain Mallaby's closing register — Hassabis as both Nobel laureate and CEO, preferring the label 'realist'", "demonstrationIds": [ "d-closing" ] } ], "demonstrations": [ { "id": "d-gemini-def", "itemIds": [ "def-gemini", "cz-infinity-machine-def-gemini", "tf-d-infinity-machine-def-gemini", "concept-rw-infinity-machine-gemini-ultra" ], "label": "Define Gemini", "requiredCorrect": 4 }, { "id": "d-mmlu", "itemIds": [ "def-mmlu", "tf-d-infinity-machine-def-mmlu", "concept-rw-infinity-machine-mmlu", "numeric-gemini-mmlu" ], "label": "MMLU & 90%", "requiredCorrect": 4 }, { "id": "d-gemini15", "itemIds": [ "q-im-gemini15-release", "q-im-gemini15-year", "concept-rw-infinity-machine-gemini-1-5-pro", "tf-f-infinity-machine-q-im-gemini15-release", "ot-cov-obj-gemini15-build-1", "ot-cov-obj-gemini15-build-2" ], "label": "Gemini 1.5 Pro", "requiredCorrect": 3 }, { "id": "d-build", "itemIds": [ "mcq-rw-l10-p1-gemini", "concept-rw-infinity-machine-sparse-mixture-of-experts" ], "label": "RL wall & MoE" }, { "id": "d-firing", "itemIds": [ "q-im-altman-firing-triggers", "tf-f-infinity-machine-q-im-altman-firing-triggers" ], "label": "Altman firing triggers" }, { "id": "d-closing", "itemIds": [ "q-im-hassabis-nobel", "tf-f-infinity-machine-q-im-hassabis-nobel", "mcq-rw-l10-p2-end" ], "label": "Hassabis the realist" } ], "objectiveTests": [ { "id": "test-gemini-define", "objectiveId": "obj-gemini-define", "title": "Gemini and Gemini Ultra", "mcqIds": [ "ot-gemini-define-1", "ot-gemini-define-2", "ot-gemini-define-3", "ot-gemini-define-4", "ot-gemini-define-5" ] }, { "id": "test-mmlu-score", "objectiveId": "obj-mmlu-score", "title": "The MMLU benchmark", "mcqIds": [ "ot-mmlu-1", "ot-mmlu-2", "ot-mmlu-3", "ot-mmlu-4", "ot-mmlu-5" ] }, { "id": "test-gemini15-build", "objectiveId": "obj-gemini15-build", "title": "Gemini 1.5 Pro and the RL wall", "mcqIds": [ "mcq-rw-l10-p1-gemini", "ot-gemini15-1", "ot-gemini15-2", "ot-gemini15-3", "ot-gemini15-4", "ot-cov-obj-gemini15-build-1", "ot-cov-obj-gemini15-build-2" ] }, { "id": "test-altman-firing-triggers", "objectiveId": "obj-altman-firing-triggers", "title": "Why the board fired Altman", "mcqIds": [ "ot-firing-1", "ot-firing-2", "ot-firing-3", "ot-firing-4", "ot-firing-5" ] }, { "id": "test-closing", "objectiveId": "obj-closing", "title": "Mallaby's closing register", "mcqIds": [ "mcq-rw-l10-p2-end", "ot-closing-1", "ot-closing-2", "ot-closing-3", "ot-closing-4" ] } ] }, { "id": "l11-quick-recall", "title": "Quick Recall — Key people and places", "order": 11, "studyGuidePath": "/packs/infinity-machine/guides/l11-quick-recall.md", "parts": [ { "id": "l11-p1-recall", "title": "People and places", "order": 1, "itemIds": [ "pair-hassabis-q", "pair-legg-q", "pair-suleyman-q", "pair-silver-q", "pair-jumper-q", "pair-thiel-q", "pair-mnih-q", "pair-altman-q", "pair-sutskever-q", "pair-buchanan-q", "pair-irving-q", "pair-bletchley-q", "mcq-rw-l11-p1-recall", "tf-t-infinity-machine-pair-hassabis-q", "tf-f-infinity-machine-pair-legg-q", "tf-t-infinity-machine-pair-suleyman-q", "tf-f-infinity-machine-pair-silver-q", "tf-f-infinity-machine-pair-jumper-q", "tf-t-infinity-machine-pair-thiel-q", "concept-rw-infinity-machine-peter-thiel", "concept-rw-infinity-machine-sam-altman", "concept-rw-infinity-machine-ilya-sutskever", "concept-rw-infinity-machine-geoffrey-irving", "concept-rw-infinity-machine-ben-buchanan", "ot-trio-1", "ot-trio-2", "ot-trio-3", "ot-trio-4", "ot-trio-5", "ot-sci-1", "ot-sci-2", "ot-sci-3", "ot-sci-4", "ot-sci-5", "ot-money-1", "ot-money-2", "ot-money-3", "ot-money-4", "ot-safety-1", "ot-safety-2", "ot-safety-3", "ot-safety-4", "ot-safety-5", "ot-cov-obj-money-rivals-1", "ot-cov-obj-money-rivals-2" ], "studyGuideAnchor": "people-and-places" } ], "objectives": [ { "id": "obj-deepmind-trio", "statement": "Recall DeepMind's founding trio — Hassabis (2024 Nobel), Legg (coined AGI), Suleyman (Microsoft AI CEO)", "demonstrationIds": [ "d-trio", "d-trio-tf" ] }, { "id": "obj-science-team", "statement": "Identify the scientists behind DeepMind's milestones — David Silver (AlphaGo), John Jumper (AlphaFold 2), Vlad Mnih (Atari DQN)", "demonstrationIds": [ "d-science", "d-science-tf" ] }, { "id": "obj-money-rivals", "statement": "Identify DeepMind's first investor Peter Thiel (the ~£400m Google acquisition) and OpenAI's protagonists Sam Altman and Ilya Sutskever", "demonstrationIds": [ "d-money", "d-openai" ] }, { "id": "obj-safety-policy", "statement": "Identify the AI safety and policy figures and the summit site — Geoffrey Irving (DeepMind→UK AISI), Ben Buchanan (Biden's AI czar), and Bletchley Park", "demonstrationIds": [ "d-safety", "d-safety-people" ] } ], "demonstrations": [ { "id": "d-trio", "itemIds": [ "pair-hassabis-q", "pair-legg-q", "pair-suleyman-q" ], "label": "Founding trio" }, { "id": "d-trio-tf", "itemIds": [ "tf-t-infinity-machine-pair-hassabis-q", "tf-f-infinity-machine-pair-legg-q", "tf-t-infinity-machine-pair-suleyman-q" ], "label": "Trio T/F" }, { "id": "d-science", "itemIds": [ "pair-silver-q", "pair-jumper-q", "pair-mnih-q" ], "label": "Milestone scientists" }, { "id": "d-science-tf", "itemIds": [ "tf-f-infinity-machine-pair-silver-q", "tf-f-infinity-machine-pair-jumper-q" ], "label": "Scientists T/F" }, { "id": "d-money", "itemIds": [ "pair-thiel-q", "tf-t-infinity-machine-pair-thiel-q", "mcq-rw-l11-p1-recall", "concept-rw-infinity-machine-peter-thiel", "ot-cov-obj-money-rivals-1", "ot-cov-obj-money-rivals-2" ], "label": "Thiel & acquisition", "requiredCorrect": 3 }, { "id": "d-openai", "itemIds": [ "pair-altman-q", "pair-sutskever-q", "concept-rw-infinity-machine-sam-altman", "concept-rw-infinity-machine-ilya-sutskever" ], "label": "OpenAI protagonists", "requiredCorrect": 3 }, { "id": "d-safety", "itemIds": [ "pair-irving-q", "pair-buchanan-q", "pair-bletchley-q" ], "label": "Safety/policy trio" }, { "id": "d-safety-people", "itemIds": [ "concept-rw-infinity-machine-geoffrey-irving", "concept-rw-infinity-machine-ben-buchanan" ], "label": "Irving & Buchanan" } ], "objectiveTests": [ { "id": "test-deepmind-trio", "objectiveId": "obj-deepmind-trio", "title": "DeepMind's Founding Trio", "mcqIds": [ "ot-trio-1", "ot-trio-2", "ot-trio-3", "ot-trio-5", "ot-trio-4" ] }, { "id": "test-science-team", "objectiveId": "obj-science-team", "title": "The Scientists Behind the Milestones", "mcqIds": [ "ot-sci-1", "ot-sci-2", "ot-sci-3", "ot-sci-5", "ot-sci-4" ] }, { "id": "test-money-rivals", "objectiveId": "obj-money-rivals", "title": "Money and Rivals", "mcqIds": [ "ot-money-1", "ot-money-2", "mcq-rw-l11-p1-recall", "ot-money-3", "ot-money-4", "ot-cov-obj-money-rivals-1", "ot-cov-obj-money-rivals-2" ] }, { "id": "test-safety-policy", "objectiveId": "obj-safety-policy", "title": "Safety, Policy, and Bletchley", "mcqIds": [ "ot-safety-1", "ot-safety-2", "ot-safety-3", "ot-safety-4", "ot-safety-5" ] } ] }, { "id": "l-key-concepts", "title": "Key Concepts", "order": 99, "parts": [ { "id": "lc-people", "title": "Key People", "order": 0, "itemIds": [ "concept-demis-hassabis", "concept-shane-legg", "concept-mustafa-suleyman", "concept-david-silver", "concept-vlad-mnih", "concept-john-jumper", "mcq-rw-lc-people", "ot-key-people-1", "ot-key-people-2", "ot-key-people-3", "ot-key-people-4" ] }, { "id": "lc-things", "title": "Key Things", "order": 1, "itemIds": [ "concept-infinity-machine-reinforcement-learning", "concept-infinity-machine-deep-learning", "concept-infinity-machine-neural-network", "concept-infinity-machine-alphago", "concept-infinity-machine-dqn-deep-q-network", "concept-infinity-machine-alphafold", "concept-infinity-machine-agi-artificial-general-intelligence", "mcq-rw-lc-things", "ot-key-techniques-1", "ot-key-techniques-2", "ot-key-techniques-3", "ot-key-techniques-4", "ot-key-techniques-5", "ot-key-systems-1", "ot-key-systems-2", "ot-key-systems-3", "ot-key-systems-4" ] }, { "id": "lc-events", "title": "Key Events", "order": 2, "itemIds": [ "concept-google-buys-deepmind", "concept-the-day-pong-clicked", "concept-singularity-summit", "concept-the-halloween-scenario", "concept-the-atari-strike-team", "concept-musk-vs-page-on-the-lawn", "mcq-rw-lc-events", "ot-key-events-1", "ot-key-events-2", "ot-key-events-3", "ot-key-events-4" ] } ], "studyGuidePath": "/packs/infinity-machine/guides/l-key-concepts.md", "objectives": [ { "id": "obj-key-people", "statement": "Recognise the key people of the book — Hassabis, Legg, Suleyman, Silver, Mnih, Jumper — including Suleyman's conflict-resolution background", "demonstrationIds": [ "d-people-1", "d-people-2" ] }, { "id": "obj-key-techniques", "statement": "Recognise the foundational AI techniques — reinforcement learning, deep learning, neural networks, and AGI", "demonstrationIds": [ "d-methods" ] }, { "id": "obj-key-systems", "statement": "Recognise DeepMind's landmark systems — AlphaGo, DQN, and AlphaFold — including what made DQN notable (learning from raw pixels)", "demonstrationIds": [ "d-systems" ] }, { "id": "obj-key-events", "statement": "Recognize the pivotal events: Google's DeepMind acquisition, the day Pong clicked, the Singularity Summit, the Atari strike team, the Musk–Page dinner.", "demonstrationIds": [ "d-events-1", "d-events-2" ] } ], "demonstrations": [ { "id": "d-people-1", "itemIds": [ "concept-demis-hassabis", "concept-shane-legg", "concept-mustafa-suleyman", "mcq-rw-lc-people" ], "label": "Founders + conflict-resolution", "requiredCorrect": 4 }, { "id": "d-people-2", "itemIds": [ "concept-david-silver", "concept-vlad-mnih", "concept-john-jumper" ], "label": "Scientists" }, { "id": "d-methods", "itemIds": [ "concept-infinity-machine-reinforcement-learning", "concept-infinity-machine-deep-learning", "concept-infinity-machine-neural-network", "concept-infinity-machine-agi-artificial-general-intelligence" ], "label": "AI techniques", "requiredCorrect": 4 }, { "id": "d-systems", "itemIds": [ "concept-infinity-machine-alphago", "concept-infinity-machine-dqn-deep-q-network", "concept-infinity-machine-alphafold", "mcq-rw-lc-things" ], "label": "Landmark systems", "requiredCorrect": 4 }, { "id": "d-events-1", "itemIds": [ "concept-google-buys-deepmind", "concept-the-day-pong-clicked", "concept-the-atari-strike-team", "mcq-rw-lc-events" ], "label": "Acquisition, Pong, Atari", "requiredCorrect": 4 }, { "id": "d-events-2", "itemIds": [ "concept-singularity-summit", "concept-the-halloween-scenario", "concept-musk-vs-page-on-the-lawn" ], "label": "Summit, Halloween, dinner" } ], "objectiveTests": [ { "id": "test-key-people", "objectiveId": "obj-key-people", "title": "Key People", "mcqIds": [ "mcq-rw-lc-people", "ot-key-people-1", "ot-key-people-2", "ot-key-people-3", "ot-key-people-4" ] }, { "id": "test-key-techniques", "objectiveId": "obj-key-techniques", "title": "Foundational AI Techniques", "mcqIds": [ "ot-key-techniques-1", "ot-key-techniques-2", "ot-key-techniques-3", "ot-key-techniques-4", "ot-key-techniques-5" ] }, { "id": "test-key-systems", "objectiveId": "obj-key-systems", "title": "Landmark Systems", "mcqIds": [ "mcq-rw-lc-things", "ot-key-systems-1", "ot-key-systems-2", "ot-key-systems-3", "ot-key-systems-4" ] }, { "id": "test-key-events", "objectiveId": "obj-key-events", "title": "Pivotal Events", "mcqIds": [ "mcq-rw-lc-events", "ot-key-events-1", "ot-key-events-2", "ot-key-events-3", "ot-key-events-4" ] } ] } ], "icon": "∞", "shortName": "Infinity Machine" }