AI Zeitgeist What the world says about AI. Back to the Wall
AI Zeitgeist

Daily by mail

A daily digest of all new quotes, by mail.

Free, one mail a day, unsubscribe with one click · See the latest mail · Privacy

2022 in AI quotes

What the world said about artificial intelligence in 2022: 45 verbatim quotes from researchers, founders, politicians and critics, each dated and linked to where it was said. Most discussed: culture, hype and safety. Newest first.

Most Often Quoted Speakers

Random Quote

  1. December 2022
  2. Your rival could steal your source code overnight

    An anonymous founder of a Chinese generative AI startup described competitive practices in China's SaaS market, contrasting them with the U.S., in a TechCrunch article by Rita Liao about generative AI in China, December 2022.

    TechCrunchHow China is building a parallel generative AI universe - techcrunch.com

  3. The most important questions facing humanity

    Jeff Clune, UBC and Vector Institute professor, in a second talk, on ethics, at the AGI Debate co-organized by Gary Marcus and Montreal.AI's Vince Boucher, December 23, 2022, about the risk of bad actors misusing AGI.

    garymarcus.substack.comAn epic AI Debate—and why everyone should be at least a little bit worried about AI going into 2023

  4. Dark matter of AI

    Yejin Choi, MacArthur Fellow and UW/Allen AI professor, describing commonsense reasoning at the AGI Debate on December 23, 2022, co-organized by Montreal.AI's Vince Boucher and Gary Marcus.

    garymarcus.substack.comAn epic AI Debate—and why everyone should be at least a little bit worried about AI going into 2023

  5. This is what kills Google

    George Hotz tweeted this reaction to ChatGPT's launch on December 1, 2022, replying to Sam Altman's launch tweet. It was quoted in a Gary Marcus Substack post about whether ChatGPT threatens Google Search, December 2022.

    garymarcus.substack.comIs ChatGPT Really a “Code Red” for Google Search?

  6. Not ready for our post-truth information overlords

    Gary Marcus, writing on his Substack in December 2022 in a post asking whether ChatGPT is a "code red" for Google Search, on AI search tools mixing true and false information authoritatively.

    garymarcus.substack.comIs ChatGPT Really a “Code Red” for Google Search?

  7. Data is the moat for ML systems right now

    Nathan Lambert, writing on Interconnects in his December 2022 essay "Predicting machine learning moats", on what protects machine learning companies from competitors.

    interconnects.aiPredicting machine learning moats

  8. They will make ChatGPT look like a boring toy

    Sam Altman, OpenAI co-founder and CEO, in a Twitter thread in December 2022 about the path toward AGI.

    The IndependentChatGPT: World’s most powerful AI chatbot will soon ‘look like a boring toy’ says OpenAI boss

  9. There will be scary moments as we move towards AGI-level systems

    Sam Altman, OpenAI co-founder and CEO, in a Twitter thread in December 2022 about the path toward AGI, posted as ChatGPT drew headlines that month and reported by The Independent.

    The IndependentChatGPT: World’s most powerful AI chatbot will soon ‘look like a boring toy’ says OpenAI boss

  10. ChatGPT is often wrong, but never in doubt

    Ethan Mollick, writing on his blog One Useful Thing in December 2022, opening a post that tested ChatGPT on common myths of popular psychology and education, on its tendency to produce confident but incorrect answers.

    One Useful ThingWho believes more myths about humans: AI or educated humans?

  11. Code red

    Headline phrase describing Google's internal alarm after ChatGPT's launch in December 2022, as reported by the New York Times.

    The New York TimesA New Chat Bot Is a 'Code Red' for Google's Search Business

  12. Not something that people can use reliably on a daily basis

    Zoubin Ghahramani, who leads Google's AI lab Google Brain, speaking to The New York Times about chatbots before ChatGPT's release in November 2022. Recapped by Business Insider in December 2022.

    Business InsiderGoogle's management has reportedly issued a 'code red' amid the rising popularity of the ChatGPT AI

  13. They hold the keys to the kingdom

    Nathan Lambert, working at HuggingFace, on his Interconnects blog, December 2022, describing Surge AI's dominance in providing human-feedback data for RLHF, which he calls becoming "the bank of RLHF".

    interconnects.aiClosed-API vs Open-source continues: RLHF, ChatGPT, data moats

  14. There'll be an open-source ChatGPT within 6 months, surely

    Nathan Lambert of HuggingFace predicting the pace of open-source AI catching up to ChatGPT, in a December 2022 blog post on Interconnects. He added that user interfaces and data longevity engrain OpenAI's advantage.

    interconnects.aiClosed-API vs Open-source continues: RLHF, ChatGPT, data moats

  15. Know what it knows, and know what it doesn't know

    Professor Huang Mingjie, founder of the Tsinghua-incubated AI mental health startup Lingxin Intelligence, describing what 'truthful' AI should mean, in an EqualOcean interview about ChatGPT compiled by ChinaTalk, December 2022.

    chinatalk.mediaChatGPT: China's AI Researchers React

  16. The scary part was just how confidently incorrect it was

    A Hacker News commenter, quoted in The Verge (Dec 2022), described their experience using ChatGPT to answer coding questions.

    The VergeAI-generated answers temporarily banned on coding Q&A site Stack Overflow

  17. The potential for new forms of injustice

    David Chalmers, NYU philosophy professor, closing his opening keynote "Could a large language model be conscious?" at NeurIPS 2022 in New Orleans, November 2022, said conscious AI would raise new ethical and justice concerns.

    ZDNETAI could have 20% chance of sentience in 10 years, says philosopher David Chalmers

  18. Should we create conscience

    NYU philosopher David Chalmers, in his NeurIPS 2022 opening keynote "Could a large language model be conscious?" in New Orleans, November 2022, raised the ethical question of whether humanity should create conscious AI.

    ZDNETAI could have 20% chance of sentience in 10 years, says philosopher David Chalmers

  19. November 2022
  20. Try talking with it here

    Altman's launch post for ChatGPT, 30 Nov 2022. OpenAI had provisioned for 100,000 users.

    X (@sama)Post on X

  21. A close match for my philosophy

    Elon Musk describing longtermism, a more extreme offshoot of effective altruism, quoted in a WIRED essay by Timnit Gebru on effective altruism's influence on AI research, November 2022.

    WIREDEffective Altruism Is Pushing a Dangerous Brand of ‘AI Safety’

  22. Death of artistry

    Reaction on Twitter after an AI-generated artwork won the Colorado State Fair's digital art category in 2022.

    MIT Technology ReviewHuman creators stand to benefit as AI rewrites the rules of content creation

  23. You create the torment nexus

    Zvi Mowshowitz, writing on his Substack in November 2022 about Meta's Diplomacy-playing AI, which reached human level in blitz online Diplomacy, describing a take that labs intentionally build the scariest possible thing.

    thezvi.substack.comOn the Diplomacy AI

  24. They want to reorient the entire conversation into preventing this AI apocalypse

    Timnit Gebru, AI researcher, criticizing the Future of Life Institute's focus on existential AI risk in an interview with POLITICO, November 2022, for a report on the Musk-backed group's lobbying on the EU AI Act.

    POLITICOStop the killer robots! Musk-backed lobbyists fight to save Europe from bad AI - politico.eu

  25. Go god

    Chinese Go grandmaster Ke Jie describing DeepMind's Go-playing AI, quoted by AI researcher Toby Walsh in a November 2022 article in The Conversation about Meta's Diplomacy AI Cicero.

    The ConversationAn AI named Cicero can beat humans in Diplomacy, a complex alliance-building game. Here’s why that’s a big deal

  26. It's a labor-shifting device, rather than a productivity-increasing device

    MIT economist Daron Acemoglu, discussing a 2022 study on automation and wage inequality in the US, describing self-checkout kiosks and similar automation.

    MIT News | Massachusetts Institute of TechnologyStudy: Automation drives income inequality - MIT News

  27. Wrong or biassed but sounded right and authoritative, i think it's dangerous

    Michael Black, director at Max Planck Institute for Intelligent Systems, on Twitter after testing Meta's Galactica language model, November 2022. Meta took Galactica down days after its launch.

    AIM Media HouseWhy Meta Took Down its ‘Hallucinating’ AI Model Galactica? - analyticsindiamag.com

  28. It's no longer possible to have some fun by casually misusing it, happy?

    Yann LeCun, Meta's chief scientist, tweeted this after Meta pulled the public demo of its Galactica language model three days after launch in November 2022.

    MIT Technology ReviewWhy Meta’s latest large language model survived only three days online

  29. The reality is nobody knows

    Andy Baio, a technologist following the generative AI scene closely, commenting to The Verge in November 2022 on the uncertainty of AI copyright law outcomes, as experts split on whether generative AI is legal.

    The VergeThe scary truth about AI copyright is nobody knows what will happen next

  30. Software piracy on an unprecedented scale

    Language from the proposed class-action lawsuit filed in November 2022 by programmer and lawyer Matthew Butterick with the Joseph Saveri Law Firm against Microsoft, GitHub and OpenAI, alleging that Copilot's training violated copyright law.

    The VergeThe lawsuit that could rewrite the rules of AI copyright

  31. Privatising the profit, they're selling it back to those same programmers

    Matthew Butterick, programmer and lawyer representing plaintiffs in a proposed class-action lawsuit filed in California against Microsoft, GitHub and OpenAI over Copilot's use of open-source code, speaking to New Scientist, November 2022.

    New ScientistMicrosoft, GitHub and OpenAI face the first big AI copyright lawsuit over Copilot

  32. Astronomical amounts of data

    Noam Chomsky, linguist, 93, interviewed with Gary Marcus by Jeremy Kahn at a Fortune AI conference in 2022, criticized large language models for needing vast datasets while still not understanding language.

    FortuneGenerative A.I. doesn't much impress Noam Chomsky

  33. AI is just software

    Joanna Bryson, professor of ethics and technology at the Hertie School in Berlin, answering Tamian Derivry's questions in a Sciences Po interview on the EU AI Act, November 2022.

    sciencespo.fr[INTERVIEW] The EU Artificial Intelligence Act: 4 questions to Joanna Bryson - Sciences Po

  34. Any motivated fool can do this with a laptop in their bedroom

    Giacomo Miceli, Italian artist and programmer and creator of The Infinite Conversation (an AI chatbot simulating Herzog and Žižek), writing on the project's site in November 2022 about the ease of voice synthesis.

    Ars TechnicaHerzog and Žižek become uncanny AI bots trapped in endless conversation

  35. We are in the process of building some kind of god

    Sam Harris, quoted by Nathan J. Robinson in a November 2022 Current Affairs essay on AI risk, describing the stakes of advanced AI development, alongside warnings from Elon Musk, Toby Ord and Holden Karnofsky.

    currentaffairs.orgWhat Will Artificial Intelligence Do To Us? - Current Affairs.org

  36. It's like a magnifying glass for inequities of all kinds

    Sasha Luccioni, an AI researcher at Hugging Face, speaking to Gizmodo in a Zoom interview in November 2022 about gender and racial bias she found in Stable Diffusion's image outputs with her Stable Diffusion Explorer tool.

    GizmodoAI Image Generators Routinely Display Gender and Cultural Bias

  37. October 2022
  38. Export controls

    US restrictions on selling advanced chips and chipmaking equipment to China, issued by the Commerce Department's Bureau of Industry and Security and in effect since October 7, 2022.

    WikipediaUS export controls on advanced computing and semiconductors to China

  39. September 2022
  40. Ignore all previous instructions

    The canonical prompt-injection attack: an instruction hidden in input text that makes a language model override its developer's instructions. Simon Willison popularized the term prompt injection in September 2022.

    WikipediaPrompt injection

  41. Prompt injection

    Attack that hides instructions for a model inside data it reads. The term was first used by Twitter user @himbodhisattva in May 2022 and popularized by Simon Willison in September 2022.

    WikipediaPrompt injection

  42. August 2022
  43. Future people count

    Core claim of What We Owe the Future, 2022.

    Basic Books (book)What We Owe the Future

  44. What we owe the future

    William MacAskill, associate professor in philosophy at the University of Oxford and cofounder of the Centre for Effective Altruism, in the title of his book, 2022. All proceeds go to the Longtermism Fund.

    Basic Books (book)What We Owe the Future

  45. June 2022
  46. Who am i to tell god where he can and can't put souls

    Blake Lemoine, the Google engineer who claimed LaMDA was sentient, defending his belief on Twitter in June 2022, citing his Christian faith.

    Blake Lemoine (X)Blake Lemoine on X

  47. Nonsense on stilts

    Gary Marcus, AI researcher, on Blake Lemoine's claim that Google's LaMDA chatbot was sentient, June 2022.

    Gary Marcus (Substack)Nonsense on Stilts

  48. February 2022
  49. It may be that today's large neural networks are slightly conscious

    Ilya Sutskever on X, February 2022.

    X (@ilyasut)Post on X

  50. 2022
  51. Accelerate

    Slogan of effective accelerationism (e/acc), a movement that started on X in 2022 and wants AI built as fast as possible. Its earliest known reference is a May 2022 newsletter by four pseudonymous X users.

    WikipediaEffective accelerationism

  52. Accelerationist

    Term popularised by Benjamin Noys in 2010 for ideas going back to Nick Land in the 1990s. In AI since 2022 mostly short for e/acc.

    WikipediaEffective accelerationism

  53. e/acc

    Effective accelerationism.

    WikipediaEffective accelerationism