AGI Zeitgeist

Panel evidence

Panel B: earliest and latest statements on AGI (checked 2026-09-30)

A working document. It is one of the documents the site is built from, published as it is written for anyone checking the work, and it uses the project's own shorthand. The plain account of the same material is on the home page.

Panel B: earliest and latest statements on AGI (checked 2026-09-30)

Inclusion bar: at least one documented public statement on whether AGI is possible, or when it arrives, ON OR BEFORE 2017, and at least one in 2024 OR LATER.

Every quote below was copied from a page fetched in this session (via WebFetch, which returns the page through a summarising model, so wording should be spot-checked against the live page before final publication). "Primary" means the person's own writing, a transcript, or their own post. "Secondhand" means a news or third-party page reporting their words.

Name Earliest statement (date, quote, link, primary/secondhand) Latest statement (date, quote, link, primary/secondhand) Clears bar Notes
Sam Altman 2015-02-25. "Development of superhuman machine intelligence (SMI) is probably the greatest threat to the continued existence of humanity." Also: "It is extremely hard to put a timeframe on when this will happen." Blog post "Machine intelligence, part 1", https://blog.samaltman.com/machine-intelligence-part-1. Primary (own blog). Date confirmed by LessWrong linkpost, https://lesswrong.com/posts/QnBZkNJNbJK9k5Xi7/linkpost-sam-altman-s-2015-blog-posts-machine-intelligence (the first quote was seen only in that linkpost and in search summaries; the second was fetched from the blog itself). July 2026 (exact day not given). "we are in the singularity" (podcast), as quoted in Gary Marcus, "Sorry, Sam and Elon, we have not reached the Singularity", 2026-07-28, https://garymarcus.substack.com/p/sorry-sam-and-elon-we-have-not-reached. Secondhand. Latest primary item: blog post written after the attack on his home, c. 2026-04-10/11: "No one understands the impacts of superintelligence yet, but they will be immense." https://blog.samaltman.com/2279512. Primary. Yes 2015 item covers both possibility and timing (says timing is hard to call; "probably still somewhat far away"). Latest dated timing claim: India AI Impact Summit, reported 2026-02-23: "On our current trajectory, we believe we may be only a couple of years away from early versions of true superintelligence." https://tech.yahoo.com/ai/articles/sam-altman-backs-democratic-ai-160058276.html (secondhand). The April 2026 post has no date printed on it; the date comes from news coverage (Business Today, 2026-04-11).
Gary Marcus January 2015 (month only). "What I think about machines thinking is that it won't happen anytime soon." Edge 2015 Annual Question response, "Machines Won't Be Thinking Anytime Soon", https://www.edge.org/response-detail/26175. Primary (own essay). 2026-09-03. "Success on ARC-AGI is great and impressive, but not—despite the name of the task—proof of AGI." "Hot take on GPT-6 Astra", https://garymarcus.substack.com/p/hot-take-on-gpt-6-astra. Primary (own newsletter). Yes Earliest item is about timing ("not anytime soon"). Earlier pieces probably exist (New Yorker columns 2012 to 2013, e.g. "Why We Should Think About the Threat of Artificial Intelligence", Oct 2013) but newyorker.com and the ACM mirror would not load, so they are unverified. Latest explicit timing claim: 2026-07-28, "If there is, it's not one that's going to happen in this decade." (about a single "Singularity" moment), https://garymarcus.substack.com/p/sorry-sam-and-elon-we-have-not-reached (primary). Also 2025-12-20: "We won't get to AGI in 2026 (or 7)." https://garymarcus.substack.com/p/six-or-seven-predictions-for-ai-2026 (primary).
Andrew Ng 2015-03-19. "I don't work on not turning AI evil today for the same reason I don't worry about the problem of overpopulation on the planet Mars." Talk at Nvidia's GPU Technology Conference, reported in The Register, https://www.theregister.com/2015/03/19/andrew_ng_baidu_ai/. Secondhand (direct-quote report of a talk). 2026-01-02. "When businesses hype up that they might achieve AGI within a few quarters, they usually try to justify these statements by setting a much lower bar." The Batch, issue 334 (his weekly letter), https://www.deeplearning.ai/the-batch/issue-334/. Primary. Yes 2015 item is about possibility and distance, framed as a far-future concern ("a race of killer robots in the far future"), with no date given. Same 2026 letter: "A reasonable definition of AGI is AI that can do any intellectual task that a human can." I only checked The Batch letters up to issue 335 (2026-01-09) for AGI; later 2026 letters were scanned by title only, so a later item may exist.
Stuart Russell 2014-04-19. "There is no physical law precluding particles from being organized in ways that perform even more advanced computations than the arrangements of particles in human brains." "Transcending Complacency on Superintelligent Machines", co-written with Stephen Hawking, Max Tegmark and Frank Wilczek, HuffPost, https://www.huffpost.com/entry/artificial-intelligence_b_5174265. Primary (co-written opinion piece). 2025-01-31. "In other words, the AGI race is a race towards the edge of a cliff." Newsweek opinion piece by Russell, "DeepSeek, OpenAI, and the Race to Human Extinction", https://www.newsweek.com/deepseek-openai-race-human-extinction-2023482. Primary (own byline). Yes The 2014 piece covers possibility, and gestures at timing: "We'll arrive in a few decades" (comparison with a message from aliens). Before 2005: his textbook Artificial Intelligence: A Modern Approach (with Peter Norvig, 1995) has a section titled "What if AI succeeds?", but I found no fetchable copy, so it is unverified. No 2026 statement was found. The search budget ran out before I could look for one, so treat "latest = Jan 2025" as a lower bound. The 2025 item treats AGI as a live race but gives no date.
Rodney Brooks 2008-06-01. "I strongly suspect it won't happen before 2030, the year that some singularitarians predict." IEEE Spectrum, "I, Rodney Brooks, Am a Robot", https://spectrum.ieee.org/i-rodney-brooks-am-a-robot. Primary (own essay). 2026-01-01. Lists as "NIML" (not in my lifetime, which he defines as not before 2050): "A robot that has any real idea about its own existence, or the existence of humans in the way that a six year old understands humans." "Predictions Scorecard, 2026 January 01", https://rodneybrooks.com/predictions-scorecard-2026-january-01/. Primary. Yes 2008 item is about timing. Same essay: "I have no idea how, exactly, this creation will come about. I also don't know when it will happen." Before 2005: Flesh and Machines (2002), unverified. The 2026 item is a prediction he first made in 2018 and re-scored in 2026; the fetch found no sentence using the words "AGI" or "superintelligence" in it. His later 2026 posts (2026-07-26 and 2026-08-01) were checked and contain no AGI statement.
Elon Musk October 2014 (MIT AeroAstro Centennial Symposium). Wikipedia's paraphrase: "During a 2014 interview at the MIT AeroAstro Centennial Symposium, Musk described AI as humanity's largest existential threat". https://en.wikipedia.org/wiki/Views_of_Elon_Musk. Secondhand (encyclopedia paraphrase, not his words). Earliest of his own words I fetched: 2017-10-05 tweet, "The singularity for this level of the simulation is coming soon." x.com/elonmusk/status/915957815197319168, as quoted in Marcus 2026-07-28 (link below). Secondhand copy of a primary post. 2026-01-04. "We have entered the Singularity" x.com/elonmusk/status/2007738847397036143, as quoted in Gary Marcus, 2026-07-28, https://garymarcus.substack.com/p/sorry-sam-and-elon-we-have-not-reached. Secondhand copy of a primary post. Yes His statements are mostly off-hand (tweets, stage remarks). The 2014 remarks ("summoning the demon" at MIT, Oct 2014; the Aug 2014 tweet calling AI "potentially more dangerous than nukes") are well known, but every news page I tried (Guardian, Washington Post, CNBC, The Verge, Business Insider, X mirror) was blocked, so I have no verbatim 2014 quote. The 2017 tweet is a timing statement ("coming soon"). Wikipedia's AI-risk article says he joined Altman in July 2026 in saying the singularity had arrived, but it gives no source link, so a later item than Jan 2026 may exist. Also 2025-02-23: "We are on the event horizon of the singularity" (same Marcus page).
Ilya Sutskever 2015-12-11. "Because of AI's surprising history, it's hard to predict when human-level AI might come within reach." "Introducing OpenAI", by Greg Brockman, Ilya Sutskever and the OpenAI team. Fetched from a verbatim copy at https://tm.durusau.net/?p=66175, because openai.com/blog/introducing-openai returned an access-denied error. Primary, co-written (copied by a third party). 2025-11-25. Asked how long until a system that learns as well as a human and then becomes superhuman: "I think like 5 to 20." (confirmed "5 to 20 years"). Dwarkesh Patel podcast, https://www.dwarkesh.com/p/ilya-sutskever-2. Primary (transcript). Yes (with a caveat: the pre-2018 item is co-written) The pre-2018 item is about TIMING: it says timing is unpredictable and assumes it will arrive ("When it does…"). Same post: "in the extreme case they will reach human performance on virtually every intellectual task." It is a jointly signed company announcement, not a solo statement. I found no solo public Sutskever statement on AGI from 2015 to 2017. His NIPS 2017 remarks were about self-play. His 2016 to 2019 film footage was not released until 2023. His 2017 emails to Musk were private and only made public in 2024. In 2026: remarks recorded for his National Academy of Sciences award (2026-05-05) talk about AI's impact, not AGI timing (seen only in search summaries, not fetched). If the panel needs a solo pre-2018 statement, mark him unclear.
François Chollet 2017-07-18. "this perpetually-learning model-growing system could be interpreted as an AGI—an Artificial General Intelligence." Also: "don't expect any singularitarian robot apocalypse to ensue: that's a pure fantasy". "The future of deep learning", Keras blog, https://blog.keras.io/the-future-of-deep-learning.html. Primary. 2026-03-27 (date of the article; date of the interview not given). "I think we're probably looking at AGI by 2030, early 2030s, most likely." OfficeChai, https://officechai.com/ai/agi-likely-by-early-2030s-when-arc-agi-6-or-7-will-be-released-arc-prizes-francois-chollet/. Secondhand. Yes (pre-2018 item is about possibility only) The pre-2018 item is about POSSIBILITY only, not timing. It says how an AGI might be built and that "intelligence explosion" fears are fantasy, with no date. His "The impossibility of intelligence explosion" (Medium, Nov 2017) also argues about possibility, but Medium returned access-denied, so it is unverified. I found no pre-2018 timing statement. Earlier timing statement, 2025: "A year ago, I would have said ten years-ish. And now I think it's probably, you know, five-ish." (talk with Dwarkesh Patel, reported 2025-08-22 by The Neuron, https://www.theneuron.ai/explainer-articles/the-clock-is-ticking-ai-skepticfrancois-chollet-slashes-agi-timeline-to-5-years-unveils-new-benchmark-to-prove-it/, secondhand).

Gaps to close in a later session

Queries run (all on 2026-09-30)

Web searches:

  1. Ilya Sutskever 2016 interview artificial general intelligence
  2. Sutskever 2017 talk "general intelligence" OpenAI self-play quote (extended)
  3. Wired 2016 Inside OpenAI Sutskever quote superintelligence
  4. "Machine intelligence, part 1" Altman February 2015
  5. "Introducing OpenAI" Brockman Sutskever "hard to predict when human-level AI might come within reach"
  6. Ilya Sutskever 2026 superintelligence interview
  7. François Chollet AGI timeline 2026
  8. Chollet ARC-AGI-3 launch 2026 "AGI" said
  9. François Chollet interview 2026 AGI years away (extended)
  10. Sam Altman 2026 blog superintelligence AGI
  11. Sam Altman Molotov cocktail blog post 2026 superintelligence
  12. Sam Altman says superintelligence by 2028 interview
  13. Gary Marcus New Yorker 2013 "Why We Should Think About the Threat of Artificial Intelligence"
  14. Gary Marcus substack AGI 2026
  15. "Six Predictions for AI in 2026" Gary Marcus
  16. Gary Marcus "AGI" garymarcus.substack.com July OR August 2026
  17. Gary Marcus 2026 "AGI is not" OR "no AGI" substack post Brundage bet update (extended)
  18. Gary Marcus 2013 New Yorker "smarter than us" machines "artificial intelligence" Barrat Our Final Invention review
  19. edge.org 2015 "machines that think" Gary Marcus response
  20. edge.org response-detail Gary Marcus "Machines Won't Be Thinking Anytime Soon"
  21. Stuart Russell Hawking Tegmark Wilczek 2014 Independent "Transcending complacency on superintelligent machines"
  22. Stuart Russell "Of Myths and Moonshine" edge.org 2014 (blocked: search budget used up)
  23. Helper agent's searches, all blocked by the budget: "Stuart Russell AGI 2026 interview superintelligence"; "Andrew Ng AGI 2026 The Batch decades away"; "Rodney Brooks predictions scorecard 2026 January 01"; "Musk 2026 AI smarter than all humans combined by year prediction"

Search pages fetched directly after the budget ran out: DuckDuckGo "Stuart Russell AGI 2026 interview" (returned a bot check); Bing "Stuart Russell AGI 2026 interview" and ""Stuart Russell" Berkeley AGI superintelligence 2026" (irrelevant results); Brave "Elon Musk Davos 2026 AI smarter than any human end of year" (rate-limited).

Pages fetched directly (beyond those cited above): Altman blog index; Marcus Substack archive and posts of 2026-09-24, 2026-09-29, 2026-09-30; The Batch index and issue 335; Rodney Brooks blog index and posts of 2026-07-26 and 2026-08-01; Wikipedia pages for Stuart J. Russell, Elon Musk, Views of Elon Musk, and Existential risk from AI; Edge 2015 question index; Edge "The Myth of AI" conversation (Russell's reply was cut off); Stanford eCorner "Inside OpenAI" (2023, not useful). Attempts that failed: openai.com, medium.com, scribe.rip, freedium, web.archive.org, newyorker.com, nytimes.com, independent.co.uk, theguardian.com, washingtonpost.com, cnbc.com, theverge.com, businessinsider.com, xcancel.com, cser.ac.uk, PBS NOVA, benzinga.com.

Rendered from docs/panel-evidence-2026-10/statements-core-b.md at version a7bd17a.