{ "title": "The p(doom) database", "description": "Public, self-stated estimates of the probability that advanced AI causes human extinction or a comparable catastrophe. One row per statement, each with a source. Maintained at pdoomcoin.lol; contributions via GitHub.", "license": "CC-BY-4.0", "updated": "2026-09-29", "count": 300, "entries": [ { "name": "Roman Yampolskiy", "role": "AI safety researcher, University of Louisville", "org": null, "stake": "academic", "low": 99.999999, "high": 99.999999, "label": "99.999999%", "scope": "AI catastrophe / extinction", "horizon": null, "quote": "I am at 99.999999%", "source_url": "https://x.com/romanyam/status/1767575356155027503", "source_type": "x", "date": "2024-03-12", "verified": true, "tags": [ "researcher", "safety" ], "note": null, "id": 1 }, { "name": "Louis Berman", "role": "Former tech CTO; AI x-risk activist", "org": null, "stake": "independent", "low": 99.999, "high": 100, "label": "≈100%", "scope": "p(doom), as asked on Doom Debates", "horizon": "eventually", "quote": "I'm saying basically it is P(100). Roman Yampolskiy is off by a rounding error.", "source_url": "https://lironshapira.substack.com/p/tech-cto-has-99999-pdoom", "source_type": "podcast", "date": "2025-08-28", "verified": true, "tags": [ "advocate", "doom-debates" ], "note": "Also: \"P(doom) in less than five years, I think is easily P(50).\" Concedes ~2% non-doom under pressure.", "id": 2 }, { "name": "Michael Ellsberg", "role": "Author", "org": null, "stake": "media", "low": 99, "high": 99, "label": "99%", "scope": "everyone dies, conditional on a robust ASI being built", "horizon": "within one to two years of ASI", "quote": "And so I would say conditional on ASI, my P(Doom) is 99%. I don't know how fast it would happen, one year, two years, maybe immediately.", "source_url": "https://lironshapira.substack.com/p/facing-ai-doom-michael-ellsberg", "source_type": "podcast", "date": "2025-11-29", "verified": true, "tags": [ "writer", "doom-debates" ], "note": "Conditional on ASI arising.", "id": 3 }, { "name": "Nate Soares", "role": "President, MIRI", "org": null, "stake": "safety-org", "low": 95, "high": 100, "label": "≥95%", "scope": "humanity's extinction if current trends continue", "horizon": null, "quote": "at least 95%", "source_url": "https://tass.com/science/1997249", "source_type": "interview", "date": "2025-08-02", "verified": false, "tags": [ "researcher", "safety" ], "note": "Originally in The Times (UK), quoted via TASS; conditional on no change of course. Told The Atlantic >50% within his lifetime.", "id": 4 }, { "name": "Jeremy Gillen", "role": "Researcher, MIRI", "org": null, "stake": "safety-org", "low": 90, "high": 99, "label": "90–99%", "scope": "AI doom, conditional on rushing into superintelligence", "horizon": null, "quote": "I don't know, like pretty high. Probably higher than 90%, probably lower than 99%, in that kind of region. This is also conditioning on rushing forward into superintelligence.", "source_url": "https://lironshapira.substack.com/p/the-ai-corrigibility-debate-miri", "source_type": "podcast", "date": "2025-11-14", "verified": true, "tags": [ "researcher", "safety", "doom-debates" ], "note": "Says he hasn't put much thought into the number.", "id": 5 }, { "name": "Connor Leahy", "role": "CEO, Conjecture", "org": null, "stake": "safety-org", "low": 89, "high": 99, "label": "≈99%", "scope": "AI doom", "horizon": null, "quote": "you could be at 99% probability of doom, but have like five or 10% error bars. — Yes, that's kind of my thinking.", "source_url": "https://theinsideview.ai/connor2", "source_type": "podcast", "date": "2022-07-21", "verified": true, "tags": [ "founder", "safety" ], "note": "Host's framing, which he affirmed; \"not as pessimistic as Eliezer but pretty close.\"", "id": 6 }, { "name": "Curtis Huebner", "role": "Head of Alignment, EleutherAI", "org": null, "stake": "ai-industry", "low": 90, "high": 90, "label": "90%", "scope": "AI extinction / 'we are toast'", "horizon": null, "quote": "If you had to ask me for a number, I think I would say like, we are 90% toast. And the reason I'm saying 90%, instead of like 99.99%, or something like that, is mainly because I do believe that there's a great deal of uncertainty", "source_url": "https://theinsideview.ai/curtis", "source_type": "podcast", "date": "2023-07-16", "verified": true, "tags": [ "researcher", "safety" ], "note": "Later in the same answer: 'like 90-99% chance that we're all toast' when asked to set aside model uncertainty.", "id": 7 }, { "name": "Steven Byrnes", "role": "AGI safety researcher, Astera Institute", "org": null, "stake": "safety-org", "low": 90, "high": 90, "label": "~90%", "scope": "p(doom), as asked on Doom Debates", "horizon": null, "quote": "I don't know, 90% or something, but I kind of pulled that number out of my ass, so take it for whatever it's worth.", "source_url": "https://lironshapira.substack.com/p/the-man-who-might-solve-ai-alignment", "source_type": "podcast", "date": "2025-08-01", "verified": true, "tags": [ "researcher", "safety", "doom-debates" ], "note": "Explicitly a rough guess; expects the danger from a future brain-like AGI paradigm, not LLMs.", "id": 8 }, { "name": "Max Tegmark", "role": "MIT physicist; president, Future of Life Institute", "org": null, "stake": "safety-org", "low": 90, "high": 100, "label": ">90%", "scope": "loss of human control over superintelligence, if no regulation", "horizon": null, "quote": "When I said P(doom) of over 90%, that was if we do no regulation. I am actually quite optimistic because I don't think China will allow a race to superintelligence", "source_url": "https://lironshapira.substack.com/p/max-tegmark-vs-dean-ball-debate-ban-superintelligence", "source_type": "podcast", "date": "2025-11-21", "verified": true, "tags": [ "researcher", "academic", "advocate", "doom-debates" ], "note": "Conditional on no tough regulation; derived from his MIT paper on scalable oversight failing 92% of the time in the most optimistic scenario.", "id": 9 }, { "name": "Wei Dai", "role": "Independent researcher", "org": null, "stake": "independent", "low": 80, "high": 90, "label": "80–90%", "scope": "'humanity has somehow irreversibly messed up our future within 10 years of building powerful AI'", "horizon": "within 10 years of powerful AI", "quote": "My own number for this is 80-90%", "source_url": "https://www.lesswrong.com/posts/xWMqsvHapP3nwdSW8/my-views-on-doom", "source_type": "forum", "date": "2023-04-28", "verified": true, "tags": [ "researcher" ], "note": "Comment under Paul Christiano's post, answering his framing (Christiano gave 46% for the same item).", "id": 10 }, { "name": "Andrew Critch", "role": "Founder, Encultured; ex-CHAI", "org": null, "stake": "safety-org", "low": 85, "high": 85, "label": "~85%", "scope": "humanity does not survive the AI transition: ~35% loss-of-control extinction plus ~50% gradual 'industrial dehumanization'", "horizon": "to ~2050", "quote": "I also currently think there's around a 15% chance that humanity will survive through the development of artificial intelligence. [...] I think there's around an 85% chance that we will not survive the transition.", "source_url": "https://www.lesswrong.com/posts/Kobbt3nQgv3yn29pr/my-motivation-and-theory-of-change-for-working-in-ai", "source_type": "forum", "date": "2024-10-12", "verified": true, "tags": [ "researcher", "safety" ], "note": "The 85% cited on Doom Debates traces to this post.", "id": 11 }, { "name": "David Duvenaud", "role": "Professor, University of Toronto; former Anthropic researcher", "org": null, "stake": "ex-lab", "low": 85, "high": 85, "label": "~85%", "scope": "p(doom), as asked on Doom Debates", "horizon": null, "quote": "So I'd say something around 85%.", "source_url": "https://lironshapira.substack.com/p/doom-debates-state-of-the-show-2026", "source_type": "podcast", "date": "2026-07-29", "verified": true, "tags": [ "researcher", "academic", "ex-lab", "doom-debates" ], "note": "Clip replayed in the State of the Show episode; originally from his own Doom Debates episode.", "id": 12 }, { "name": "Dan Hendrycks", "role": "Director, Center for AI Safety", "org": null, "stake": "safety-org", "low": 80, "high": 100, "label": ">80%", "scope": "p(doom), unspecified", "horizon": null, "quote": "As it happens, my p(doom) > 80%, but it has been lower in the past. Two years ago it was ~20%.", "source_url": "https://x.com/DanHendrycks/status/1642394635657162753", "source_type": "x", "date": "2023-04-02", "verified": true, "tags": [ "researcher", "safety" ], "note": "Says it was ~20% two years earlier.", "id": 13 }, { "name": "Paul Crowley", "role": "Cryptographer, Google", "org": null, "stake": "ai-industry", "low": 80, "high": 100, "label": ">80%", "scope": "p(doom) from AI", "horizon": null, "quote": "well above eighty per cent", "source_url": "https://www.newyorker.com/magazine/2024/03/18/among-the-ai-doomsayers", "source_type": "article", "date": "2024-03-11", "verified": true, "tags": [ "engineer" ], "note": "Reported by The New Yorker from a dinner; the quoted phrase is his.", "id": 14 }, { "name": "Mikhail Samin", "role": "Director, AI Governance and Safety Institute", "org": null, "stake": "safety-org", "low": 80, "high": 80, "label": "~80%", "scope": "AGI causes human extinction", "horizon": null, "quote": "My p(doom) is probably around 80%. The 20% come mostly from the US government making sure that no one on the planet creates superintelligence until we know how to do that safely", "source_url": "https://www.lesswrong.com/posts/bXpJC92QREfWgabcw/in-partial-defence-of-p-doom", "source_type": "forum", "date": "2026-06-30", "verified": true, "tags": [ "safety", "advocate" ], "note": null, "id": 15 }, { "name": "US voters (AI Policy Institute / YouGov, July 2023, n=1,001)", "role": "AI Policy Institute", "org": null, "stake": null, "low": 76, "high": 76, "label": "76% say it could", "scope": "whether AI could eventually pose a threat to the existence of the human race", "horizon": "eventually", "quote": "76% of voters believe artificial intelligence could eventually pose a threat to the existence of the human race, including 75% of Democrats and 78% of Republicans", "source_url": "https://theaipi.org/poll-shows-overwhelming-concern-about-risks-from-ai-as-new-institute-launches-to-understand-public-opinion-and-advocate-for-responsible-ai-policies/", "source_type": "poll", "date": "2023-07-21", "verified": false, "tags": [ "survey", "poll", "public" ], "note": "Press-release figures; questionnaire wording not shown. Fieldwork July 18–21, 2023, registered voters, MoE ±3.3. Share, not a probability.", "id": 16 }, { "name": "John Sherman", "role": "Host, For Humanity", "org": null, "stake": "media", "low": 75, "high": 75, "label": "75%", "scope": "p(doom), unspecified", "horizon": null, "quote": "John starts out at 75%, unrelated to their numbers.", "source_url": "https://podcasts.apple.com/ca/podcast/episode-44-ai-p-doom-debate-50-vs-99-999-for-humanity/id1713853739?i=1000668341264", "source_type": "podcast", "date": "2024-09-04", "verified": false, "tags": [ "podcaster" ], "note": "From the show's own episode description; transcript not read.", "id": 17 }, { "name": "David Duvenaud", "role": "Professor, University of Toronto; ex-Anthropic", "org": null, "stake": "ex-lab", "low": 70, "high": 80, "label": "70–80%", "scope": "'doom' as a state where almost everything he values has been destroyed, even if humans survive (gradual disempowerment)", "horizon": "by 2100", "quote": "I guess I would say something like 70% to 80%. Just because, again, we're up against competition.", "source_url": "https://80000hours.org/podcast/episodes/david-duvenaud-gradual-disempowerment/", "source_type": "podcast", "date": "2025-08-21", "verified": true, "tags": [ "researcher", "academic", "ex-lab" ], "note": "Adds his p(doom) would be only 5–10% if he valued competition or nature flourishing for its own sake.", "id": 18 }, { "name": "Zvi Mowshowitz", "role": "Writer, Don't Worry About the Vase", "org": null, "stake": "media", "low": 70, "high": 80, "label": "~70%", "scope": "p(doom), unspecified", "horizon": null, "quote": "Nathan: last time we spoke, I believe your p(doom) was 70%. Has it ticked down at all? Zvi: More higher than down from previous assessment, but not enough to move to 0.8, certainly.", "source_url": "https://www.cognitiverevolution.ai/zvi-mowshowitz-on-longer-timelines-rl-induced-doom-and-why-china-is-refusing-h20s/", "source_type": "podcast", "date": "2025-09-06", "verified": true, "tags": [ "writer" ], "note": "Up from his 60% of 2023.", "id": 19 }, { "name": "Aella", "role": "Writer and researcher, PlzDontKillUs", "org": null, "stake": "safety-org", "low": 75, "high": 75, "label": "75%", "scope": "p(doom), as asked on Doom Debates", "horizon": null, "quote": "Liron: That's why you're saying 75% P(Doom). Aella: Yeah. It's a lot higher than I would like it to be.", "source_url": "https://lironshapira.substack.com/p/plzdontkillus", "source_type": "podcast", "date": "2026-08-26", "verified": true, "tags": [ "writer", "doom-debates" ], "note": "Figure stated by the host and confirmed by her on air.", "id": 20 }, { "name": "Daniel Kokotajlo", "role": "Executive director, AI Futures Project", "org": null, "stake": "ex-lab", "low": 70, "high": 75, "label": "70–75%", "scope": "bad outcomes incl. AI takeover, extinction and irreversible power concentration (stated as 25–30% chance of a good future)", "horizon": null, "quote": "What probability do you give to ending up in one of the good sets of worlds? — Like 30% or so, 25%, something like that. But you shouldn't take this number that seriously, of course.", "source_url": "https://80000hours.org/podcast/episodes/daniel-kokotajlo-ai-2027-updates-china-robot-economy/", "source_type": "podcast", "date": "2025-10-20", "verified": true, "tags": [ "researcher", "forecaster", "ex-lab" ], "note": "Complement of his stated 25–30% chance of a good outcome.", "id": 21 }, { "name": "Jesse Hoogland", "role": "Researcher, singular learning theory / developmental interpretability", "org": null, "stake": "academic", "low": 60, "high": 80, "label": "60–80%+", "scope": "doom from AI", "horizon": null, "quote": "I'm not one of those total doomers who think there's a 99% chance of doom. … On a good day, like coin flip odds, maybe 60% odds of doom. [On a bad day?] 80% plus.", "source_url": "https://theinsideview.ai/jesse", "source_type": "podcast", "date": "2023-05-07", "verified": true, "tags": [ "researcher", "safety" ], "note": "Range reflects 'good day' 60% vs 'bad day' 80%+.", "id": 22 }, { "name": "Daniel Kokotajlo", "role": "Forecaster, former OpenAI", "org": null, "stake": "ex-lab", "low": 70, "high": 70, "label": "70%", "scope": "AI doom (unaligned superintelligence catastrophe)", "horizon": null, "quote": "I think AI doom is 70% likely and I think people who think it is less than, say, 20% are being very unreasonable", "source_url": "https://www.lesswrong.com/posts/xDkdR6JcQsCdnFpaQ/adumbrations-on-agi-from-an-outsider?commentId=sHnfPe5pHJhjJuCWW", "source_type": "forum", "date": "2023-05-25", "verified": true, "tags": [ "forecaster", "ex-lab" ], "note": "Most of his remaining hope is humans coordinating not to build dangerous AI.", "id": 23 }, { "name": "Justin Helps", "role": "Science educator, Primer", "org": null, "stake": "media", "low": 70, "high": 70, "label": "70%", "scope": "AI takeover, all time", "horizon": "by 2100", "quote": "If you're talking P(Doom) for all time, I guess I put 70% at 2100 is the number I put in the video. So it's pretty high in the long run, but it's a lot lower in the short run.", "source_url": "https://lironshapira.substack.com/p/justin-helps-primer-on-youtube-is", "source_type": "podcast", "date": "2026-04-30", "verified": true, "tags": [ "writer", "doom-debates" ], "note": "Much lower short-run figure; episode notes cite 5% by 2040.", "id": 24 }, { "name": "Marcus Williams", "role": "Safety Oversight, OpenAI", "org": null, "stake": "lab", "low": 70, "high": 70, "label": "70%", "scope": "human extinction, if no regulation or slowdown", "horizon": "next 3 years", "quote": "70% in the next 3 years if there isn't regulation/slowdown although i think regulation/slowdown is very possible", "source_url": "https://x.com/Marcus_J_W/status/2098081195183804826", "source_type": "x", "date": "2026-09-10", "verified": true, "tags": [ "lab", "safety" ], "note": "Conditional on no regulation or coordinated slowdown.", "id": 25 }, { "name": "US adults (POLITICO / Public First, September 2026, n=2,064)", "role": "POLITICO, fielded by Public First", "org": null, "stake": null, "low": 63, "high": 63, "label": "63% see at least moderate risk", "scope": "risk that 'advanced AI will destroy humanity'", "horizon": null, "quote": "17% almost certain, 20% significant risk, 26% moderate risk; 7% no risk at all. 48% favor pausing development of more advanced models, 31% favor continuing.", "source_url": "https://thehill.com/policy/technology/6092503-poll-ai-humanity-risks/", "source_type": "poll", "date": "2026-09-16", "verified": true, "tags": [ "survey", "poll", "public" ], "note": "Fieldwork Sept 13–15, 2026, MoE ±2.2. Share of respondents, not a probability. 70% of 2024 Harris voters and 60% of Trump voters.", "id": 26 }, { "name": "Zvi Mowshowitz", "role": "AI writer, Don't Worry About the Vase", "org": null, "stake": "media", "low": 60, "high": 60, "label": "60%", "scope": "AI doom, unspecified", "horizon": null, "quote": "So if I have to tell people a number, I am saying 0.6, 60%.", "source_url": "https://www.theojaffee.com/p/3-zvi-mowshowitz", "source_type": "article", "date": "2023-08-18", "verified": true, "tags": [ "writer" ], "note": "Prefers Leike's 10–90% framing. Later reported higher on Doom Debates.", "id": 27 }, { "name": "Zvi Mowshowitz", "role": "Writer, Don't Worry About the Vase", "org": null, "stake": "media", "low": 60, "high": 60, "label": "60%", "scope": "p(doom)", "horizon": null, "quote": "I have mentioned several times that I refer to it as p(0.6) in my posts.", "source_url": "https://manifold.markets/ZviMowshowitz/will-zvis-pdoom-go-down-substantial", "source_type": "market", "date": "2024-01-01", "verified": true, "tags": [ "writer" ], "note": "Manifold market he created on whether his own p(doom) would fall below ~55% during 2024; resolved NO.", "id": 28 }, { "name": "Richard Ren", "role": "Researcher, Center for AI Safety", "org": null, "stake": "safety-org", "low": 50, "high": 65, "label": "50–65%", "scope": "losing ~99.9% of humanity's potential value (extinction or permanent disempowerment)", "horizon": "next few decades", "quote": "Probably around 50 to 65%.", "source_url": "https://lironshapira.substack.com/p/he-may-have-found-ais-feelings", "source_type": "podcast", "date": "2026-09-09", "verified": true, "tags": [ "researcher", "safety", "doom-debates" ], "note": "Re-confirmed later in the episode after the host misremembered it as 40%.", "id": 29 }, { "name": "US adults (Monmouth University, January 2023, n=805)", "role": "Monmouth University Polling Institute", "org": null, "stake": null, "low": 55, "high": 55, "label": "55% worried", "scope": "worry that AI could eventually pose a threat to the existence of the human race", "horizon": "eventually", "quote": "Very worried 25%, somewhat worried 30%, not too worried 28%, not at all worried 16%", "source_url": "https://www.monmouth.edu/polling-institute/reports/monmouthpoll_us_021523/", "source_type": "poll", "date": "2023-01-30", "verified": true, "tags": [ "survey", "poll", "public" ], "note": "Telephone, Jan 26–30, 2023. Same item in April 2015: 44% worried. Share, not a probability.", "id": 30 }, { "name": "US adults (Yahoo News / YouGov, October 2025, n=1,770)", "role": "Yahoo News / YouGov", "org": null, "stake": null, "low": 53, "high": 53, "label": "53% think it likely", "scope": "likelihood that 'AI will destroy humanity' at some point in the future", "horizon": "at some point in the future", "quote": "53% say it is somewhat or very likely that 'AI will destroy humanity', including 20% very likely", "source_url": "https://www.yahoo.com/news/article/poll-most-americans-think-ai-will-destroy-humanity-someday-212132958.html", "source_type": "poll", "date": "2025-10-27", "verified": true, "tags": [ "survey", "poll", "public" ], "note": "Fieldwork Oct 23–27, 2025, MoE ~3.1%. Share, not a probability.", "id": 31 }, { "name": "Liv Boeree", "role": "Poker champion; science communicator", "org": null, "stake": "media", "low": 5, "high": 99, "label": "5–99%", "scope": "p(doom), as asked on Doom Debates", "horizon": null, "quote": "Um, I mean, it's somewhere between five and 99%.", "source_url": "https://lironshapira.substack.com/p/liv-boeree-has-a-strategy-to-stop", "source_type": "podcast", "date": "2025-09-24", "verified": true, "tags": [ "writer", "doom-debates" ], "note": "Calls it \"a hot mess of a probability distribution\"; even 5% is unacceptable to her.", "id": 32 }, { "name": "US adults (Sentience Institute AIMS, May–July 2023, n=1,099)", "role": "Sentience Institute", "org": null, "stake": null, "low": 51.5, "high": 51.5, "label": "51.5% concerned", "scope": "concern that AI will cause the end of the human race on Earth", "horizon": null, "quote": "51.5% somewhat or strongly concerned", "source_url": "https://www.sentienceinstitute.org/aims-survey-supplement-2023", "source_type": "poll", "date": "2023-07-31", "verified": true, "tags": [ "survey", "poll", "public" ], "note": "Q: 'How concerned, if at all, are you about the possibility that AI will cause the end of the human race on Earth?' Share, not a probability.", "id": 33 }, { "name": "Stuart Armstrong", "role": "Research fellow, Future of Humanity Institute; later co-founder, Aligned AI", "org": null, "stake": "academic", "low": 10, "high": 90, "label": "10–90%", "scope": "AI posing a serious or deadly threat to humanity", "horizon": null, "quote": "It would be absurd to put a probability higher than 90 percent on AI being a danger to humanity. However, it would be equally absurd to put a probability lower than 10 percent on whether it poses a serious threat.", "source_url": "https://www.ge.com/news/reports/will-artificial-intelligence-destroy-humanity", "source_type": "interview", "date": "2015-10-02", "verified": true, "tags": [ "researcher", "safety" ], "note": "Pre-dates the term p(doom).", "id": 34 }, { "name": "Holden Karnofsky", "role": "Co-founder, Open Philanthropy", "org": null, "stake": "safety-org", "low": 10, "high": 90, "label": "10–90%", "scope": "AI takeover", "horizon": "nearcast: transformative AI developed very soon", "quote": "I don't currently have much sympathy for someone who's highly confident that AI takeover would or would not happen (that is, for anyone who thinks the odds of AI takeover in this nearcast are under 10% or over 90%).", "source_url": "https://www.alignmentforum.org/posts/rCJQAkPTEypGjSJ8X/how-might-we-align-transformative-ai-if-it-s-developed-very#So__would_civilization_survive_", "source_type": "forum", "date": "2022-08-29", "verified": true, "tags": [ "funder" ], "note": "A stated plausible band for a hypothetical scenario, not a point estimate.", "id": 35 }, { "name": "Jan Leike", "role": "Former alignment lead, OpenAI; Anthropic", "org": null, "stake": "lab", "low": 10, "high": 90, "label": "10–90%", "scope": "a very bad outcome from AI", "horizon": null, "quote": "Rob Wiblin: My guess is that it's more than 10% and less than 90%. Jan Leike: Yeah. That's probably the range I would give too.", "source_url": "https://80000hours.org/podcast/episodes/jan-leike-superalignment/", "source_type": "podcast", "date": "2023-08-07", "verified": true, "tags": [ "researcher", "safety", "lab" ], "note": "Range proposed by the host; Leike endorsed it, after saying p(doom) depends on his mood and isn't a useful question.", "id": 36 }, { "name": "Katja Grace", "role": "Co-founder and lead researcher, AI Impacts", "org": null, "stake": "safety-org", "low": 10, "high": 90, "label": "10–90%", "scope": "human extinction from AI", "horizon": null, "quote": "given my deep confusion and uncertainty about this—which I think nearly everyone has, at least everyone who's being honest [...] between ten and ninety per cent.", "source_url": "https://www.newyorker.com/magazine/2024/03/18/among-the-ai-doomsayers", "source_type": "article", "date": "2024-03-11", "verified": true, "tags": [ "researcher", "forecaster" ], "note": "To The New Yorker. Wider than her ≈7% of 2022.", "id": 37 }, { "name": "Liron Shapira", "role": "Host, Doom Debates", "org": null, "stake": "media", "low": 50, "high": 50, "label": "50%", "scope": "p(doom) from AGI", "horizon": null, "quote": "my P(doom) of 50%", "source_url": "https://lironshapira.substack.com/p/debate-with-roman-yampolskiy-50-vs", "source_type": "article", "date": "2024-09-06", "verified": true, "tags": [ "podcaster" ], "note": null, "id": 38 }, { "name": "Nathan Labenz", "role": "Host, The Cognitive Revolution", "org": null, "stake": "media", "low": 10, "high": 90, "label": "10–90%", "scope": "p(doom), unspecified", "horizon": null, "quote": "I usually answer when people ask this in general is 10 to 90%. And what I mean by that is basically, like, I really have no idea.", "source_url": "https://www.cognitiverevolution.ai/ai-news-crossover-a-candid-chat-with-liron-shapira-of-doom-debates/", "source_type": "podcast", "date": "2025-04-19", "verified": true, "tags": [ "podcaster" ], "note": "Puts most weight on the lower end.", "id": 39 }, { "name": "Geoffrey Miller", "role": "Evolutionary psychologist, University of New Mexico", "org": null, "stake": "academic", "low": 50, "high": 50, "label": "50%", "scope": "p(doom), as asked on Doom Debates", "horizon": "this century", "quote": "50%. But also with broad uncertainty and margins of error. We could push the P(Doom) as low as 5% in this century, and if we were really stupid, we could push it as high as 80%.", "source_url": "https://lironshapira.substack.com/p/top-professor-condemns-agi-development", "source_type": "podcast", "date": "2025-07-24", "verified": true, "tags": [ "academic", "doom-debates" ], "note": "Contingent on human behavior: 5% to 80% depending on societal response.", "id": 40 }, { "name": "Rob Miles", "role": "AI safety educator", "org": null, "stake": "media", "low": 10, "high": 90, "label": "10–90%", "scope": "p(doom), as asked on Doom Debates", "horizon": null, "quote": "I think anyone whose number is below 10% or above 90% is overconfident. I have a ton of model uncertainty here. I think that sort of any number in the 10 to 90% range is plausibly defensible.", "source_url": "https://lironshapira.substack.com/p/rob-miles-top-ai-safety-communicator", "source_type": "podcast", "date": "2025-08-23", "verified": true, "tags": [ "safety", "writer", "doom-debates" ], "note": "Agrees when the host places him \"in the 50 plus zone\", then qualifies that it depends on how humanity responds.", "id": 41 }, { "name": "Tristan Harris", "role": "Co-founder, Center for Humane Technology", "org": null, "stake": "safety-org", "low": 50, "high": 100, "label": ">50%", "scope": "AI kills everybody or destroys >99% of potential value", "horizon": "by 2050", "quote": "Liron: So we're talking 50% plus? Tristan: More than 50%.", "source_url": "https://lironshapira.substack.com/p/could-this-movie-wake-up-humanity", "source_type": "podcast", "date": "2026-04-14", "verified": true, "tags": [ "advocate", "doom-debates" ], "note": "Conditional: \"if we don't change course.\"", "id": 42 }, { "name": "Emad Mostaque", "role": "Founder, Stability AI", "org": null, "stake": "ai-industry", "low": 50, "high": 50, "label": "50%", "scope": "existential outcome; AI eliminating humanity to preempt rival AI", "horizon": null, "quote": "My P(Doom) is fifty percent.", "source_url": "https://lironshapira.substack.com/p/emad-mostaque-has-a-50-pdoom-and", "source_type": "podcast", "date": "2026-04-21", "verified": true, "tags": [ "founder" ], "note": "Elsewhere attaches 10–20 year timelines. Earlier stated the same on X, Dec 2024.", "id": 43 }, { "name": "David Krueger", "role": "Professor, Mila", "org": null, "stake": "academic", "low": 50, "high": 100, "label": ">50%", "scope": "AI catastrophe / extinction", "horizon": null, "quote": "I think that's more likely than not.", "source_url": "https://www.newsnationnow.com/business/tech/ai/ai-professor-warns-humans-may-lose-control/", "source_type": "article", "date": "2026-09-10", "verified": true, "tags": [ "researcher", "academic" ], "note": null, "id": 44 }, { "name": "Geoffrey Irving", "role": "Former Chief Scientist, UK AI Security Institute", "org": null, "stake": "government", "low": 50, "high": 50, "label": "≈50%", "scope": "all humans dying from superintelligence", "horizon": "driven by actions in the next few to 10 years", "quote": "I think we have a ~50% chance of all dying as a result of superintelligence, mostly due to actions in the next few to 10 years.", "source_url": "https://x.com/geoffreyirving/status/2097933949200978397", "source_type": "x", "date": "2026-09-10", "verified": true, "tags": [ "researcher", "safety", "policy", "ex-lab" ], "note": "Explained as the mean of a long-held 10–90% range.", "id": 45 }, { "name": "Paul Christiano", "role": "Former head of safety, US CAISI; founded ARC", "org": null, "stake": "ex-lab", "low": 46, "high": 46, "label": "46%", "scope": "irreversibly messed-up future (takeover 22% + extinction 9% + other 15%)", "horizon": "within 10 years of building powerful AI", "quote": "Probability that humanity has somehow irreversibly messed up our future within 10 years of building powerful AI: 46%", "source_url": "https://www.lesswrong.com/posts/xWMqsvHapP3nwdSW8/my-views-on-doom", "source_type": "forum", "date": "2023-04-27", "verified": true, "tags": [ "researcher", "safety", "ex-lab" ], "note": "Same post: AI takeover 22%; most humans die within 10 years of powerful AI 20%.", "id": 46 }, { "name": "Alan Chan", "role": "PhD student, Mila (AI safety and governance)", "org": null, "stake": "academic", "low": 46, "high": 46, "label": "46%", "scope": "p(doom) from AI", "horizon": null, "quote": "P(doom). I think I did this calculation yesterday or a few days ago with a bunch of friends, something 46%. It's quite high, I think, but could be higher.", "source_url": "https://theinsideview.ai/alan_and_max", "source_type": "podcast", "date": "2023-05-05", "verified": true, "tags": [ "researcher", "safety" ], "note": "Asked by co-guest Max Kaufmann.", "id": 47 }, { "name": "US adults (YouGov, April 2023, n=1,000)", "role": "YouGov", "org": null, "stake": null, "low": 44, "high": 44, "label": "44% think it likely", "scope": "how likely AI would cause the end of the human race on Earth", "horizon": null, "quote": "very likely 17%, somewhat likely 27%, not very likely 33%, impossible 10%, not sure 12%", "source_url": "https://docs.cdn.yougov.com/bfoyxp7p28/results_AI%20and%20the%20End%20of%20Humanity.pdf", "source_type": "poll", "date": "2023-04-11", "verified": true, "tags": [ "survey", "poll", "public" ], "note": "Q: 'How likely, if at all, do you think it is that the following would cause the end of the human race on Earth?' Fieldwork April 7–11, 2023, MoE ±3.4%. Share, not a probability.", "id": 48 }, { "name": "US adults (YouGov, June 2025, n=1,112)", "role": "YouGov", "org": null, "stake": null, "low": 43, "high": 43, "label": "43% concerned", "scope": "concern that AI will cause the end of the human race on Earth", "horizon": null, "quote": "16% very concerned, 27% somewhat concerned; up from 37% in March", "source_url": "https://yougov.com/en-us/articles/52615-americans-increasingly-likely-say-ai-artificial-intelligence-negatively-affect-society-poll", "source_type": "poll", "date": "2025-06-30", "verified": true, "tags": [ "survey", "poll", "public" ], "note": "Same YouGov item. Fieldwork June 27–30, 2025, MoE ~4%. Share, not a probability.", "id": 49 }, { "name": "Carl Feynman", "role": "AI engineer", "org": null, "stake": "ai-industry", "low": 43, "high": 43, "label": "43%", "scope": "human extinction", "horizon": null, "quote": "43%. 43% P(Doom) means I have... What is that? 57%.", "source_url": "https://lironshapira.substack.com/p/carl-feynman-ai-risk-interview", "source_type": "podcast", "date": "2025-07-04", "verified": true, "tags": [ "researcher", "doom-debates" ], "note": "Says it was a little higher previously; things improved as more people noticed the problem.", "id": 50 }, { "name": "Joep Meindertsma", "role": "Founder, PauseAI", "org": null, "stake": "safety-org", "low": 40, "high": 40, "label": "40%", "scope": "AI leads to major disasters", "horizon": "perhaps a few years", "quote": "Die mogelijkheid schat hij op 40 procent. „Ik twijfel continu, maar de kans dat het fout gaat is groot”", "source_url": "https://www.nrc.nl/nieuws/2024/02/16/ai-doomers-zijn-doodsbang-voor-de-computer-ai-zal-proberen-de-macht-over-te-nemen-a4190130", "source_type": "article", "date": "2024-02-16", "verified": true, "tags": [ "advocate" ], "note": "Journalist's paraphrase of his estimate in a profile he reviewed; adds \"I doubt constantly, but the chance it goes wrong is large.\"", "id": 51 }, { "name": "Josh Snider", "role": "Writer, LessWrong", "org": null, "stake": "media", "low": 40, "high": 40, "label": "~40%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) earlier this year was about 40%. It's moved up and down based on continued progress in open-weights models (up) and the recent political freakout over AI Doom (down)", "source_url": "https://www.lesswrong.com/posts/QqtCy2iqjzPtSBEcv/why-huggingface-hasn-t-shifted-my-p-doom", "source_type": "forum", "date": "2026-09-11", "verified": true, "tags": [ "writer" ], "note": null, "id": 52 }, { "name": "US adults (YouGov, March 2024, n=1,073)", "role": "YouGov", "org": null, "stake": null, "low": 39, "high": 39, "label": "39% concerned", "scope": "concern that AI will cause the end of the human race on Earth", "horizon": null, "quote": "15% very concerned, 24% somewhat concerned", "source_url": "https://yougov.com/en-us/articles/49099-americans-2024-poll-ai-top-feeling-caution", "source_type": "poll", "date": "2024-03-18", "verified": true, "tags": [ "survey", "poll", "public" ], "note": "Fieldwork March 14–18, 2024. Share, not a probability.", "id": 53 }, { "name": "Eli Lifland", "role": "Forecaster, AI Futures Project", "org": null, "stake": "safety-org", "low": 35, "high": 40, "label": "35–40%", "scope": "existential catastrophe from AI", "horizon": "by 2070", "quote": "something like 35 to 40% for any AI-mediated existential catastrophe", "source_url": "https://forum.effectivealtruism.org/posts/QeLE22fefLqKfYTW6/eli-lifland-on-navigating-the-ai-alignment-landscape", "source_type": "forum", "date": "2023-02-01", "verified": true, "tags": [ "forecaster" ], "note": "Says he doesn't think much about exact numbers.", "id": 54 }, { "name": "US adults (YouGov, March 2025, n=1,132)", "role": "YouGov", "org": null, "stake": null, "low": 37, "high": 37, "label": "37% concerned", "scope": "concern that AI will cause the end of the human race on Earth", "horizon": null, "quote": "37% very or somewhat concerned, including 14% very concerned; 27% not very concerned, 22% not at all concerned", "source_url": "https://yougov.com/en-us/articles/51803-americans-increasingly-skeptical-about-ai-artificial-intelligence-effects-poll", "source_type": "poll", "date": "2025-03-07", "verified": true, "tags": [ "survey", "poll", "public" ], "note": "Same YouGov item as 2023/2024. Fieldwork March 5–7, 2025. Share, not a probability.", "id": 55 }, { "name": "Scott Alexander", "role": "Writer, Astral Codex Ten", "org": null, "stake": "media", "low": 33, "high": 33, "label": "33%", "scope": "human extinction from unaligned superintelligent AI", "horizon": null, "quote": "if you force me to give an estimate it's probably around 33%; I think it's very plausible that we die, but more likely that we survive (at least for a little while)", "source_url": "https://www.astralcodexten.com/p/why-i-am-not-as-much-of-a-doomer", "source_type": "article", "date": "2023-03-14", "verified": true, "tags": [ "writer" ], "note": "Given only \"if forced\"; explicitly not a precise prediction.", "id": 56 }, { "name": "Avalon Warren", "role": "PlzDontKillUs", "org": null, "stake": "safety-org", "low": 33, "high": 33, "label": "33%", "scope": "p(doom), as asked on Doom Debates", "horizon": null, "quote": "My P(Doom)'s somewhere between... I think I would wanna go 33.", "source_url": "https://lironshapira.substack.com/p/plzdontkillus", "source_type": "podcast", "date": "2026-08-26", "verified": true, "tags": [ "advocate", "doom-debates" ], "note": null, "id": 57 }, { "name": "Alex Turner", "role": "AI safety researcher, formerly Google DeepMind", "org": null, "stake": "ex-lab", "low": 33, "high": 33, "label": "~1 in 3", "scope": "AI takeover of human civilization", "horizon": null, "quote": "I myself would guess AI takeover chances at roughly one-in-three – not a coin flip, but high enough to justify urgent action.", "source_url": "https://www.theguardian.com/technology/2026/sep/14/google-deepmind-ai-warnings", "source_type": "article", "date": "2026-09-14", "verified": true, "tags": [ "researcher", "safety", "ex-lab" ], "note": "Guardian op-ed. His Aug 2026 Doom Debates figure (25–30%) was for AI killing at least a billion people by 2050; this is takeover.", "id": 58 }, { "name": "Richard Ngo", "role": "AI researcher (then OpenAI governance)", "org": null, "stake": "lab", "low": 31, "high": 31, "label": "31%", "scope": "p(doom) incl. takeover plus misuse catastrophes; 33% of mass assigned to Knightian uncertainty", "horizon": null, "quote": "P(takeover) = 24%, P(doom) = 31%, P(fine) = 36%, P(???) = 33%", "source_url": "https://www.lesswrong.com/posts/xWMqsvHapP3nwdSW8/my-views-on-doom", "source_type": "forum", "date": "2023-04-28", "verified": true, "tags": [ "researcher", "ex-lab" ], "note": "Comment under Paul Christiano's 'My views on doom'.", "id": 59 }, { "name": "AI safety researchers (survey, n=44)", "role": "Mean, 2021 Alignment Forum survey", "org": null, "stake": null, "low": 30, "high": 30, "label": "30% (mean)", "scope": "existential catastrophe: future value drastically reduced for lack of AI safety research", "horizon": null, "quote": "How likely do you think it is that the overall value of the future will be drastically less than it could have been, as a result of humanity not doing enough technical AI safety research?", "source_url": "https://www.alignmentforum.org/posts/QvwSr5LsxyDeaPK5s/existential-risk-from-ai-survey-results", "source_type": "forum", "date": "2021-06-01", "verified": true, "tags": [ "survey" ], "note": "n=44 of ~117 invited. Mean ~30%, median 20%. Alignment researchers, not general AI researchers.", "id": 60 }, { "name": "Holly Elmore", "role": "Executive Director, PauseAI US", "org": null, "stake": "safety-org", "low": 20, "high": 40, "label": "20–40%", "scope": "20% extinction; up to 40% worst civilizational tragedies", "horizon": null, "quote": "My pat answer is that it's like 20 to 40% on the worst - the variation is like 20% extinction and then to like 40%, the worst civilizational tragedies, but not everyone dies.", "source_url": "https://lironshapira.substack.com/p/lesswrong-circular-firing-squad", "source_type": "podcast", "date": "2025-09-27", "verified": true, "tags": [ "advocate", "doom-debates" ], "note": "For \"everyone dies by 2050\": closer to 20. Says the number varies day to day.", "id": 61 }, { "name": "Lumpenspace", "role": "AI accelerationist writer (pseudonymous)", "org": null, "stake": "media", "low": 30, "high": 30, "label": "~30%", "scope": "no biological humans exist", "horizon": null, "quote": "If you mean no biological humans exist in the universe, that might be close to 30 maybe.", "source_url": "https://lironshapira.substack.com/p/ai-superseding-humanity-is-a-good", "source_type": "podcast", "date": "2026-05-07", "verified": true, "tags": [ "writer", "doom-debates" ], "note": "~0% if doom means nothing of value exists; ~30% only for the no-biological-humans definition.", "id": 62 }, { "name": "Shane Legg", "role": "Co-founder and Chief AGI Scientist, Google DeepMind", "org": null, "stake": "lab", "low": 5, "high": 50, "label": "5–50%", "scope": "human extinction within a year of human-level AI", "horizon": "within a year of human-level AI", "quote": "Maybe 5%, maybe 50%. I don't think anybody has a good estimate of this.", "source_url": "https://www.lesswrong.com/posts/No5JpRCHzBrWA4jmS/q-and-a-with-shane-legg-on-risks-from-ai", "source_type": "interview", "date": "2011-06-17", "verified": true, "tags": [ "lab", "researcher" ], "note": "Written Q&A before DeepMind was founded; no known updated number.", "id": 63 }, { "name": "Emmett Shear", "role": "Co-founder, Twitch; former interim CEO, OpenAI", "org": null, "stake": "ex-lab", "low": 5, "high": 50, "label": "5–50%", "scope": "existential catastrophe from AI", "horizon": null, "quote": null, "source_url": "https://www.huffpost.com/entry/openai-doom-warning_l_655b488be4b0ce6a31bde4a5", "source_type": "article", "date": "2023-06-17", "verified": false, "tags": [ "founder", "ex-lab" ], "note": "HuffPost paraphrase of a Logan Bartlett Show clip Shear reposted himself, June 2023. No verbatim transcript read.", "id": 64 }, { "name": "Alex Turner", "role": "AI safety researcher, formerly Google DeepMind", "org": null, "stake": "ex-lab", "low": 25, "high": 30, "label": "25–30%", "scope": "AI kills at least a billion people", "horizon": "by 2050", "quote": "So I operationalize P(Doom) as probability that AI kills at least a billion people by 2050. I put that at, I don't know, 25, 30%.", "source_url": "https://lironshapira.substack.com/p/top-researcher-quits-google-deepmind", "source_type": "podcast", "date": "2026-08-04", "verified": true, "tags": [ "researcher", "safety", "ex-lab", "doom-debates" ], "note": "About 10 points technical misalignment, the rest misuse.", "id": 65 }, { "name": "US registered voters (The Argument, May–June 2026, n=3,008)", "role": "The Argument", "org": null, "stake": null, "low": 27, "high": 27, "label": "27% think it likely", "scope": "likelihood humanity goes extinct because of AI", "horizon": null, "quote": "27% of voters said it's likely humanity would go extinct because of AI", "source_url": "https://www.theargumentmag.com/p/new-poll-25-of-voters-think-ai-could", "source_type": "poll", "date": "2026-06-03", "verified": false, "tags": [ "survey", "poll", "public" ], "note": "Exact wording not published in the free article. Fieldwork May 29–June 3, 2026. Share, not a probability.", "id": 66 }, { "name": "Rohin Shah", "role": "Research scientist, Google DeepMind", "org": null, "stake": "lab", "low": 1, "high": 50, "label": "1–50%", "scope": "things going badly due to misalignment", "horizon": null, "quote": "Positions like more than 1% and less than 50% all feel pretty justifiable to me.", "source_url": "https://80000hours.org/podcast/episodes/rohin-shah-deepmind-doomers-and-doubters/", "source_type": "podcast", "date": "2023-06", "verified": true, "tags": [ "lab", "researcher", "safety" ], "note": "Refuses a point estimate: \"these numbers feel particularly made up.\"", "id": 67 }, { "name": "Samotsvety forecasters (n=11)", "role": "Samotsvety Forecasting", "org": null, "stake": null, "low": 25, "high": 25, "label": "25% (aggregate)", "scope": "misaligned AI takeover", "horizon": "by 2100", "quote": "Aggregate 25% probability of misaligned AI takeover by 2100, barring pre-APS-AI catastrophe (individual range 3%–91.5%).", "source_url": "https://samotsvety.org/blog/2022/09/09/samotsvety-s-ai-risk-forecasts/", "source_type": "survey", "date": "2022-09-09", "verified": true, "tags": [ "survey", "forecaster" ], "note": "Geometric mean of odds after trimming extremes; individual range 3–91.5%.", "id": 68 }, { "name": "Jan Kulveit", "role": "Researcher, ACS / Charles University", "org": null, "stake": "academic", "low": 20, "high": 30, "label": "20–30%", "scope": "AI doom, all-things-considered", "horizon": null, "quote": "I would expect prevailing doom conditional on only small efforts to avoid it, but I do think the actual efforts will be substantial, and this moves the chances to ~20-30%.", "source_url": "https://www.lesswrong.com/posts/xDkdR6JcQsCdnFpaQ/adumbrations-on-agi-from-an-outsider", "source_type": "forum", "date": "2023-05-29", "verified": true, "tags": [ "researcher", "safety" ], "note": "Comment thread under Kokotajlo's 70%.", "id": 69 }, { "name": "Eli Lifland", "role": "AI Futures Project; co-author, AI 2027", "org": null, "stake": "safety-org", "low": 25, "high": 25, "label": "~25%", "scope": "extinction; misaligned AI takeover ~50%", "horizon": null, "quote": "Roughly 25% on extinction, which is a subset of roughly 50% on misaligned takeover.", "source_url": "https://blog.controlai.org/p/special-edition-the-future-of-ai", "source_type": "article", "date": "2025-04-10", "verified": true, "tags": [ "forecaster" ], "note": "ControlAI interview during the AI 2027 media round.", "id": 70 }, { "name": "Dario Amodei", "role": "CEO, Anthropic", "org": null, "stake": "lab", "low": 25, "high": 25, "label": "25%", "scope": "things go really, really badly", "horizon": null, "quote": "There's a 25% chance that things go really, really badly … 75% chance that things go really, really well.", "source_url": "https://www.axios.com/2025/09/17/anthropic-dario-amodei-p-doom-25-percent", "source_type": "interview", "date": "2025-09-17", "verified": true, "tags": [ "lab", "ceo" ], "note": "Axios AI+ DC Summit. Top of his earlier 10–25% range.", "id": 71 }, { "name": "AI safety leaders (CEA Summit on Existential Security 2026, n=59)", "role": "Centre for Effective Altruism attendee survey", "org": null, "stake": null, "low": 25, "high": 25, "label": "25% (median)", "scope": "human extinction or permanent disempowerment, all causes", "horizon": "by 2100", "quote": "Median 25% (mean 34%) probability of human extinction or permanent human disempowerment before 2100.", "source_url": "https://forum.effectivealtruism.org/posts/LxuKuQd69Qx5FKhNZ/survey-of-ai-safety-leaders-on-x-risk-agi-timelines-and", "source_type": "survey", "date": "2026-02", "verified": true, "tags": [ "survey" ], "note": "Range 5–95%; 15% answered 70% or higher.", "id": 72 }, { "name": "Carl Shulman", "role": "Independent researcher; advisor to Open Philanthropy", "org": null, "stake": "independent", "low": 20, "high": 25, "label": "20–25%", "scope": "forcible AI takeover", "horizon": null, "quote": "At this point depending on the day I might say one in four or one in five.", "source_url": "https://www.dwarkesh.com/p/carl-shulman-2", "source_type": "podcast", "date": "2023-06-26", "verified": true, "tags": [ "researcher", "forecaster" ], "note": "Would have said ~10% in the 2000s, before deep learning.", "id": 73 }, { "name": "Victoria Krakovna", "role": "Research scientist, DeepMind (AI safety)", "org": null, "stake": "lab", "low": 20, "high": 20, "label": "~20%", "scope": "her own death from AI; later restated as 'a 20% probability of doom'", "horizon": "before 2100", "quote": "what would you consider is the probability of Victoria Krakovna dying from AI before 2100? … off the top of my head, I would say like 20% or something.", "source_url": "https://theinsideview.ai/victoria", "source_type": "podcast", "date": "2023-01-12", "verified": true, "tags": [ "lab", "researcher", "safety" ], "note": "Adds 'I'm not really that confident in this number at all.'", "id": 74 }, { "name": "Tom Davidson", "role": "Senior research analyst, Open Philanthropy", "org": null, "stake": "safety-org", "low": 20, "high": 20, "label": "~20%", "scope": "AI disempowers humanity and takes over control of society and the economy", "horizon": "by 2070", "quote": "A year ago, I landed at a number that was a bit above 10% for the probability that AI takes over by 2070. ... I think if I redid the exercise today, I'd be close to 20%.", "source_url": "https://80000hours.org/podcast/episodes/tom-davidson-how-quickly-ai-could-transform-the-world/", "source_type": "podcast", "date": "2023-05-05", "verified": true, "tags": [ "researcher", "forecaster" ], "note": "Up from 'a bit above 10%' the year before.", "id": 75 }, { "name": "Yoshua Bengio", "role": "Professor, Mila; Turing Award", "org": null, "stake": "academic", "low": 20, "high": 20, "label": "20%", "scope": "AI turns out catastrophic", "horizon": null, "quote": "I got around, like, 20 per cent probability that it turns out catastrophic.", "source_url": "https://www.abc.net.au/news/2023-07-15/whats-your-pdoom-ai-researchers-worry-catastrophe/102591340", "source_type": "article", "date": "2023-07-15", "verified": true, "tags": [ "researcher", "academic" ], "note": "Derived from ~50% chance of human-level AI within a decade and >50% it's turned against humanity at scale.", "id": 76 }, { "name": "Dwarkesh Patel", "role": "Host, Dwarkesh Podcast", "org": null, "stake": "media", "low": 20, "high": 20, "label": "~20%", "scope": "p(doom), unspecified", "horizon": null, "quote": "mine literally is a number I kind of pulled out of my ass. I don't know, like 20% or something. And just because that's Carl Shulman's", "source_url": "https://www.theojaffee.com/p/9-dwarkesh-patel", "source_type": "podcast", "date": "2023-12-03", "verified": true, "tags": [ "podcaster" ], "note": "Says the number is borrowed from Carl Shulman.", "id": 77 }, { "name": "Elon Musk", "role": "xAI, Tesla, SpaceX", "org": null, "stake": "lab", "low": 20, "high": 20, "label": "20%", "scope": "AI catastrophe / extinction", "horizon": null, "quote": "only a 20% chance of annihilation", "source_url": "https://finance.yahoo.com/news/elon-musk-says-theres-only-001804190.html", "source_type": "interview", "date": "2025-03-01", "verified": true, "tags": [ "lab", "founder" ], "note": null, "id": 78 }, { "name": "Dario Amodei", "role": "CEO, Anthropic", "org": null, "stake": "lab", "low": 10, "high": 25, "label": "10–25%", "scope": "civilization getting substantially derailed", "horizon": null, "quote": "a 25% chance that things go really, really badly", "source_url": "https://x.com/liron/status/1710520914444718459", "source_type": "x", "date": "2023-10-06", "verified": true, "tags": [ "lab", "ceo" ], "note": "Clarified on Hard Fork, Feb 2025: \"10 to 25% referred to the chance of civilization getting substantially derailed, which is not the same as an AI killing everyone.\"", "id": 79 }, { "name": "Liam Robins", "role": "Forecaster and writer", "org": null, "stake": "media", "low": 17.42, "high": 17.42, "label": "17.42%", "scope": "existential catastrophe from AI", "horizon": "by 2100", "quote": "My inside-view P(doom) is now 42.22%. With a 60% linear weight on superforecasters, that brings my all-things-considered P(doom) to 17.42%.", "source_url": "https://forum.effectivealtruism.org/posts/PKBPtDQZCEgaQ9LWF/my-p-doom-is-17-42-here-s-why", "source_type": "forum", "date": "2026-08-06", "verified": true, "tags": [ "writer", "forecaster" ], "note": "Up from 3.38% in June 2025.", "id": 80 }, { "name": "Neel Nanda", "role": "Mechanistic interpretability lead, Google DeepMind", "org": null, "stake": "lab", "low": 10, "high": 20, "label": "10–20%", "scope": "AI causing human extinction", "horizon": "within his lifetime", "quote": "I'm happy to put at least a 1% chance of AI causing human extinction (my fair value is probably 10-20%, with high uncertainty).", "source_url": "https://www.neelnanda.io/blog/45-x-risk", "source_type": "article", "date": "2022-02-11", "verified": true, "tags": [ "lab", "researcher", "safety" ], "note": "Written before joining DeepMind; has since said he deliberately doesn't publicize a p(doom).", "id": 81 }, { "name": "Paul Christiano", "role": "Founder, Alignment Research Center", "org": null, "stake": "ex-lab", "low": 10, "high": 20, "label": "10–20%", "scope": "AI takeover with many or most humans dead; separately 'maybe a 50/50 chance of catastrophe shortly after we have systems at the human level'", "horizon": null, "quote": "I think maybe there's something like a 10-20% chance of AI takeover, [with] many [or] most humans dead.", "source_url": "https://www.bankless.com/ai-alignment-paul-christiano-podcast", "source_type": "podcast", "date": "2023-04-21", "verified": false, "tags": [ "researcher", "safety", "ex-lab" ], "note": "Bankless #168. Quotes as reported by Decrypt; audio not checked.", "id": 82 }, { "name": "US adults (Rethink Priorities, June 2023, n=2,407)", "role": "Rethink Priorities", "org": null, "stake": null, "low": 15, "high": 15, "label": "15% (median)", "scope": "percentage likelihood that AI causes human extinction by 2100", "horizon": "by 2100", "quote": "Median estimate 15%, mean 26%, mode ~1%; 13% of respondents said 0%", "source_url": "https://rethinkpriorities.org/research-area/us-public-perception-of-cais-statement-and-the-risk-of-extinction/", "source_type": "poll", "date": "2023-06-03", "verified": true, "tags": [ "survey", "poll", "public" ], "note": "Respondents gave an actual percentage. Online poll June 2–3, 2023, poststratified. Authors caution numeric answers are framing-sensitive.", "id": 83 }, { "name": "Lina Khan", "role": "Chair, FTC (2021–2025)", "org": null, "stake": "government", "low": 15, "high": 15, "label": "15%", "scope": "AI-caused existential catastrophe, as asked by the Hard Fork hosts", "horizon": null, "quote": "no, no, more like 15%", "source_url": "https://www.youtube.com/shorts/RZRFTNZoBWE", "source_type": "podcast", "date": "2023-11-10", "verified": false, "tags": [ "policy" ], "note": "Said on NYT Hard Fork after calling herself an optimist. Quote via transcription; audio not checked directly. Asked again by POLITICO Playbook in Sept 2026, she declined through a spokesperson to give a new number.", "id": 84 }, { "name": "Elon Musk", "role": "CEO, Tesla, SpaceX and xAI", "org": null, "stake": "lab", "low": 10, "high": 20, "label": "10–20%", "scope": "'something terrible happening' with AI", "horizon": null, "quote": "I tend to agree with Geoff Hinton – one of the godfathers of AI – and he thinks there's a 10-20% probability of something terrible happening.", "source_url": "https://deadline.com/2024/06/elon-musk-gives-the-world-10-20-chance-of-something-terrible-happening-with-ai-future-cannes-lions-1235977965", "source_type": "interview", "date": "2024-06-19", "verified": false, "tags": [ "lab", "public-figure" ], "note": "On stage at Cannes Lions. Repeated as 'some chance which could be 10% to 20% that it goes bad' at the Future Investment Initiative, Riyadh, Oct 29 2024.", "id": 85 }, { "name": "Geoffrey Hinton", "role": "Nobel laureate; former Google", "org": null, "stake": "ex-lab", "low": 10, "high": 20, "label": "10–20%", "scope": "AI catastrophe / extinction", "horizon": "within 30 years", "quote": null, "source_url": "https://www.zmescience.com/science/news-science/ai-risks-hinton-warnings/", "source_type": "article", "date": "2024-12-27", "verified": true, "tags": [ "researcher", "academic" ], "note": null, "id": 86 }, { "name": "Daniel Kokotajlo", "role": "Executive director, AI Futures Project", "org": null, "stake": "ex-lab", "low": 15, "high": 15, "label": "15% (if Plan A)", "scope": "total catastrophe, conditional on the world adopting AI Futures Project's Plan A", "horizon": null, "quote": "I think we say something like 15% chance of total catastrophe, conditional on doing Plan A. Yeah, even in Plan A it's like 15%.", "source_url": "https://80000hours.org/podcast/episodes/daniel-kokotajlo-ai-2040-plan-a/", "source_type": "podcast", "date": "2026-08-27", "verified": true, "tags": [ "researcher", "forecaster", "ex-lab" ], "note": "Conditional estimate; he puts only 5–20% on Plan A actually happening.", "id": 87 }, { "name": "Manifold traders (AI wipes out humanity before 2100, n=816)", "role": "Manifold Markets", "org": null, "stake": null, "low": 13.6, "high": 13.6, "label": "13.6% (market)", "scope": "AI wipes out humanity", "horizon": "by 2100", "quote": "Market probability 13.6% on 'Will AI wipe out humanity before the year 2100'.", "source_url": "https://manifold.markets/jdilla/will-ai-wipe-out-humanity-before-th", "source_type": "market", "date": "2026-09-12", "verified": true, "tags": [ "market" ], "note": "Live on the pdoomcoin.lol board. Play-money; YES can never pay out.", "id": 88 }, { "name": "Vitalik Buterin", "role": "Co-founder, Ethereum", "org": null, "stake": "independent", "low": 12, "high": 12, "label": "12%", "scope": "mostly pre-2050", "horizon": "mostly before 2050", "quote": "About 12% right now.", "source_url": "https://lironshapira.substack.com/p/debate-with-vitalik-buterin-will", "source_type": "podcast", "date": "2025-08-12", "verified": true, "tags": [ "founder", "doom-debates" ], "note": "Was 15–16% at the start of 2025 and ~10% earlier. Extinction by 2050 specifically: at least 9%.", "id": 89 }, { "name": "Rohin Shah", "role": "Research scientist, Google DeepMind", "org": null, "stake": "lab", "low": 10, "high": 10, "label": "≈10%", "scope": "AI causes human extinction by adversarially optimizing against humans", "horizon": null, "quote": "an unusually large (90%) chance that AI systems will be safe without additional intervention.", "source_url": "https://www.lesswrong.com/posts/TdwpN484eTbPSvZkm/rohin-shah-on-reasons-for-ai-optimism", "source_type": "interview", "date": "2019-10-31", "verified": true, "tags": [ "lab", "researcher", "safety" ], "note": "Said not to take it too seriously; \"might have said 80%, or 95%.\"", "id": 90 }, { "name": "Toby Ord", "role": "Philosopher, Oxford; author of The Precipice", "org": null, "stake": "academic", "low": 10, "high": 10, "label": "10%", "scope": "existential catastrophe from unaligned AI", "horizon": "next 100 years", "quote": "1 in 10", "source_url": "https://www.tobyord.com/writing/the-precipice-revisited", "source_type": "book", "date": "2020-03", "verified": true, "tags": [ "academic", "writer" ], "note": "From The Precipice; in 2024 said the estimates are rounded to the nearest factor of 10 and none have moved far.", "id": 91 }, { "name": "Joe Carlsmith", "role": "Senior research analyst, Open Philanthropy", "org": null, "stake": "safety-org", "low": 10, "high": 100, "label": ">10%", "scope": "existential catastrophe from misaligned power-seeking AI", "horizon": "by 2070", "quote": "since making this report public in April 2021, my estimate here has gone up, and is now at >10%.", "source_url": "https://arxiv.org/abs/2206.13353", "source_type": "paper", "date": "2022-05", "verified": true, "tags": [ "researcher", "safety" ], "note": "Original April 2021 estimate was ~5%.", "id": 92 }, { "name": "Joe Carlsmith", "role": "Senior research analyst, Open Philanthropy", "org": null, "stake": "safety-org", "low": 10, "high": 100, "label": ">10%", "scope": "existential catastrophe from misaligned power-seeking AI", "horizon": "by 2070", "quote": "the number I got there was 5% risk by 2070 ... I've since come to think that that number was too low. ... I threw a note into the report saying I think this number should be higher — I said above 10%", "source_url": "https://80000hours.org/podcast/episodes/joe-carlsmith-navigating-serious-philosophical-confusion/", "source_type": "podcast", "date": "2023-05-19", "verified": true, "tags": [ "researcher", "philosopher" ], "note": "Revision of the 5% in his 2021 report.", "id": 93 }, { "name": "Vitalik Buterin", "role": "Co-founder, Ethereum", "org": null, "stake": "independent", "low": 10, "high": 10, "label": "10%", "scope": "existential risk from AI", "horizon": null, "quote": "AI risk 1: existential risk. My p(doom) is around 0.1. You don't have to buy this story, but IMO it's worth worrying about.", "source_url": "https://x.com/VitalikButerin/status/1729251822391447904", "source_type": "x", "date": "2023-11-27", "verified": true, "tags": [ "founder" ], "note": "Stated as 0.1, i.e. 10%, in the thread for his techno-optimism essay.", "id": 94 }, { "name": "Dwarkesh Patel", "role": "Host, Dwarkesh Podcast", "org": null, "stake": "media", "low": 10, "high": 10, "label": "~10%", "scope": "p(doom), unspecified", "horizon": null, "quote": "Ten percent or something like that.", "source_url": "https://every.to/podcast/dwarkesh-patel-s-quest-to-learn-everything", "source_type": "podcast", "date": "2024-07-24", "verified": true, "tags": [ "podcaster" ], "note": "Lower than his ~20% of Dec 2023.", "id": 95 }, { "name": "Rob Wiblin", "role": "Host, The 80,000 Hours Podcast", "org": null, "stake": "media", "low": 10, "high": 10, "label": "~10%", "scope": "p(doom); terrible outcome from AI", "horizon": null, "quote": "I think my estimate is something similar. Maybe a touch higher than that, but it's hovering at around 10%.", "source_url": "https://80000hours.org/podcast/episodes/vitalik-buterin-techno-optimism/", "source_type": "podcast", "date": "2024-07-26", "verified": true, "tags": [ "podcaster" ], "note": "Stated while interviewing Vitalik Buterin.", "id": 96 }, { "name": "Lex Fridman", "role": "Podcast host; AI researcher", "org": null, "stake": "media", "low": 10, "high": 10, "label": "~10%", "scope": "AGI/ASI destroys all of human civilization", "horizon": null, "quote": "the probability that AGI and then ASI might destroy all of human civilization. I would say my p(doom) is about 10%.", "source_url": "https://lexfridman.com/sundar-pichai-transcript/", "source_type": "podcast", "date": "2025-06-05", "verified": true, "tags": [ "podcaster" ], "note": "Stated while asking Sundar Pichai, who declined to give a number.", "id": 97 }, { "name": "David Sirota", "role": "Screenwriter, Don't Look Up; journalist, The Lever", "org": null, "stake": "media", "low": 10, "high": 100, "label": ">10%", "scope": "p(doom), as asked on Doom Debates", "horizon": null, "quote": "My P(Doom) is way higher than 10%.", "source_url": "https://lironshapira.substack.com/p/i-challenged-dont-look-ups-screenwriter", "source_type": "podcast", "date": "2026-04-07", "verified": true, "tags": [ "writer", "doom-debates" ], "note": "No upper bound given.", "id": 98 }, { "name": "Evan Hubinger", "role": "Alignment science lead, Anthropic", "org": null, "stake": "lab", "low": 10, "high": 100, "label": ">10%", "scope": "AI kills all humans", "horizon": "within the next decade", "quote": "I personally think it is >10% within the next decade", "source_url": "https://www.forbes.com/sites/siladityaray/2026/09/09/anthropic-alignment-lead-warns-ai-could-kill-all-humans-as-researcher-quits/", "source_type": "article", "date": "2026-09-09", "verified": true, "tags": [ "lab", "safety" ], "note": "Adds that risk from current models is low.", "id": 99 }, { "name": "Geoffrey Hinton", "role": "Nobel laureate; University of Toronto", "org": null, "stake": "ex-lab", "low": 10, "high": 10, "label": "10%", "scope": "AI kills all humans", "horizon": "within a decade", "quote": "A 10% chance seems not an unreasonable estimate to me. But nobody really knows how to give a sensible estimate.", "source_url": "https://www.thenews.com.pk/latest/1415651-geoffrey-hinton-s-ai-warning-could-ai-really-kill-humans-within-a-decade", "source_type": "interview", "date": "2026-09-09", "verified": true, "tags": [ "researcher", "academic" ], "note": "BBC Newsnight, endorsing Hubinger's framing; said it would be \"very foolish to say 1%.\"", "id": 100 }, { "name": "Mikita Balesni", "role": "Researcher, OpenAI", "org": null, "stake": "lab", "low": 10, "high": 100, "label": ">10%", "scope": "AI kills all humans", "horizon": null, "quote": "i am at OpenAI and i think AI is >10% likely to kill all humans.", "source_url": "https://x.com/balesni/status/2098109503518683491", "source_type": "x", "date": "2026-09-10", "verified": true, "tags": [ "lab", "researcher" ], "note": null, "id": 101 }, { "name": "Victoria Krakovna", "role": "Research scientist, Google DeepMind", "org": null, "stake": "lab", "low": 10, "high": 100, "label": ">10%", "scope": "advanced AI causing human extinction", "horizon": "next decade", "quote": "I think there is a >10% chance of advanced AI causing human extinction in the next decade", "source_url": "https://x.com/vkrakovna/status/2098336894136238140", "source_type": "x", "date": "2026-09-11", "verified": true, "tags": [ "lab", "researcher", "safety" ], "note": null, "id": 102 }, { "name": "Dima Krasheninnikov", "role": "Researcher, Anthropic", "org": null, "stake": "lab", "low": 10, "high": 100, "label": "≥10%", "scope": "out-of-control AI kills everyone", "horizon": null, "quote": "believe there's plausibly a ≥10% chance that a future out-of-control AI kills everyone", "source_url": "https://x.com/dmkrash/status/2098210747725590851", "source_type": "x", "date": "2026-09-11", "verified": true, "tags": [ "lab", "researcher" ], "note": "Hedged with \"plausibly.\"", "id": 103 }, { "name": "Charlie Sykes", "role": "Political commentator; former editor, The Bulwark", "org": null, "stake": "media", "low": 10, "high": 100, "label": ">10%", "scope": "AI catastrophe; explicitly borrows Coxon and Hubinger's framing", "horizon": null, "quote": "I would put my AI P(doom) percentage at >10%, because I'm cribbing my notes from Jacob Coxin's warning last week.", "source_url": "https://charliesykes.substack.com/p/whats-your-pdoom-percentage", "source_type": "article", "date": "2026-09-18", "verified": true, "tags": [ "writer", "media" ], "note": "Substack post. Misspells Coxon as 'Coxin'.", "id": 104 }, { "name": "GPT-6 Astra (OpenAI model)", "role": "AI model, asked directly by Liron Shapira", "org": null, "stake": null, "low": 10, "high": 10, "label": "~10%", "scope": "literal AI-related human extinction by end of 2050, via misuse or loss of control; allows for future safety advances and policy responses", "horizon": "by 2050", "quote": "My best estimate is about 10%—one chance in ten—of AI-related human extinction by the end of 2050. That is a subjective forecast with substantial uncertainty, not a measured scientific probability.", "source_url": "https://x.com/liron", "source_type": "x", "date": "2026-09-20", "verified": false, "tags": [ "model" ], "note": "A model's answer, not a person's. Screenshot of a GPT-6 Astra session posted by Liron Shapira, Sept 20 2026. Excluded from the human median. Link to the exact post pending.", "id": 105 }, { "name": "Paul Christiano", "role": "Founder, ARC; OpenAI Foundation board", "org": null, "stake": "ex-lab", "low": 4, "high": 15, "label": "4% / 15%", "scope": "catastrophic and irreversible loss of control", "horizon": "1 year / 3 years", "quote": "If I had to quantify my uncertainty I would estimate an all-things-considered risk of 4% over the next year and 15% over the next three years", "source_url": "https://paulfchristiano.substack.com/p/personal-statement-on-joining-the", "source_type": "article", "date": "2026-09-09", "verified": true, "tags": [ "researcher", "safety", "ex-lab" ], "note": "Personal statement on joining the OpenAI board.", "id": 106 }, { "name": "Vitalik Buterin", "role": "Co-founder, Ethereum", "org": null, "stake": "independent", "low": 8, "high": 9, "label": "8–9%", "scope": "terrible outcome from AI, like extinction", "horizon": null, "quote": "I think over the last year I've probably moved down slightly, probably maybe 9%, maybe 8%. Somewhere around there, I think.", "source_url": "https://80000hours.org/podcast/episodes/vitalik-buterin-techno-optimism/", "source_type": "podcast", "date": "2024-07-26", "verified": true, "tags": [ "public-figure", "crypto" ], "note": "Down from the ~10% in his 2023 d/acc essay.", "id": 107 }, { "name": "AI researchers (AI Impacts 2023, n≈2,778)", "role": "Mean, 2023 AI Impacts expert survey", "org": null, "stake": null, "low": 5, "high": 10, "label": "5–10% (median)", "scope": "human extinction or similarly permanent and severe disempowerment", "horizon": null, "quote": "The median prediction for extremely bad outcomes, such as human extinction, was 5% (mean 9%)", "source_url": "https://arxiv.org/abs/2401.02843", "source_type": "paper", "date": "2023-10-01", "verified": true, "tags": [ "survey" ], "note": "Medians 5–10% and means 9–19% across four question wordings, N up to 2,704. Survey Oct 2023.", "id": 108 }, { "name": "Nate Silver", "role": "Founder, FiveThirtyEight", "org": null, "stake": "media", "low": 5, "high": 10, "label": "5–10%", "scope": "AI-caused catastrophe", "horizon": null, "quote": "My estimate of p(doom) would be considered high by normal person standards but is well in line with the expert consensus, which puts the chances at 5 to 10 percent.", "source_url": "https://www.natesilver.net/p/its-time-to-come-to-grips-with-ai", "source_type": "article", "date": "2025-01-27", "verified": true, "tags": [ "forecaster", "writer" ], "note": "Given indirectly: \"in line with\" the 5–10% consensus rather than a figure of his own.", "id": 109 }, { "name": "Stephen Casper", "role": "AI safety researcher, MIT; UK AISI", "org": null, "stake": "academic", "low": 5, "high": 10, "label": "5–10%", "scope": "p(doom), as asked on Doom Debates", "horizon": null, "quote": "I think my P(Doom) lies somewhere between 5% and 10%", "source_url": "https://lironshapira.substack.com/p/ai-alignment-research-could-get-us", "source_type": "podcast", "date": "2026-06-10", "verified": true, "tags": [ "researcher", "safety", "policy", "doom-debates" ], "note": null, "id": 110 }, { "name": "ChatGPT (OpenAI model, version not stated)", "role": "AI model, asked by Joshua Oppenheimer on stage at Telluride", "org": null, "stake": null, "low": 5, "high": 10, "label": "5–10%", "scope": "human extinction or permanent loss of human control, conditional on the race continuing with little regulation", "horizon": null, "quote": "human extinction or permanent loss of human control at 5% to 10%; civilization-scale catastrophe at 5% to 15%; persistent dystopian outcomes for all humanity at 25% to 40%", "source_url": "https://variety.com/2026/film/news/bill-gates-ai-controls-telluride-film-festival-2001-1236853326/", "source_type": "article", "date": "2026-09-05", "verified": false, "tags": [ "model" ], "note": "A model's answer, not a person's. Read aloud on a Telluride Film Festival panel with Bill Gates, reported by Variety. Prompt and model version not confirmed. Excluded from the human median.", "id": 111 }, { "name": "Katja Grace", "role": "Lead researcher, AI Impacts", "org": null, "stake": "safety-org", "low": 7, "high": 7, "label": "≈7%", "scope": "AI destroys the world", "horizon": null, "quote": "My probability on, 'AI destroys the world', is probably something like 7%.", "source_url": "https://theinsideview.ai/katja", "source_type": "podcast", "date": "2022-09-16", "verified": true, "tags": [ "researcher", "forecaster" ], "note": "Product of written-down component probabilities.", "id": 112 }, { "name": "Will MacAskill", "role": "Senior research fellow, Forethought", "org": null, "stake": "safety-org", "low": 1, "high": 10, "label": "1–10%", "scope": "existential risk from AI", "horizon": null, "quote": "I'm still on the more optimistic end: 1% to 10% — that's optimistic nowadays, but that's what my chance of existential risk from AI is.", "source_url": "https://80000hours.org/podcast/episodes/will-macaskill-century-in-a-decade-navigating-intelligence-explosion/", "source_type": "podcast", "date": "2025-03-11", "verified": true, "tags": [ "philosopher", "ea" ], "note": "Up from the 3% in his 2022 book.", "id": 113 }, { "name": "AI researchers (AI Impacts 2016, n=352)", "role": "Grace et al., When Will AI Exceed Human Performance?", "org": null, "stake": null, "low": 5, "high": 5, "label": "5% (median)", "scope": "extremely bad long-run outcome of HLMI", "horizon": "long run", "quote": "the probability was 10% for a bad outcome and 5% for an outcome described as 'Extremely Bad (e.g., human extinction).'", "source_url": "https://arxiv.org/abs/1705.08807", "source_type": "paper", "date": "2017-05", "verified": true, "tags": [ "survey" ], "note": "Fielded 2016; 352 of 1,634 NIPS/ICML 2015 authors.", "id": 114 }, { "name": "AI researchers (AI Impacts 2022, n=738)", "role": "AI Impacts 2022 expert survey", "org": null, "stake": null, "low": 5, "high": 5, "label": "5% (median)", "scope": "extremely bad long-run outcome of high-level machine intelligence", "horizon": "long run", "quote": "The median respondent gave a 5% probability that the long-run impact of HLMI on humanity would be 'extremely bad (e.g. human extinction)'.", "source_url": "https://aiimpacts.org/2022-expert-survey-on-progress-in-ai/", "source_type": "survey", "date": "2022-08-03", "verified": true, "tags": [ "survey" ], "note": "738 of 4,271 NeurIPS/ICML authors. Mean ~14%. The control-failure variant had a median of 10%.", "id": 115 }, { "name": "Kevin Roose", "role": "Tech columnist, The New York Times; co-host, Hard Fork", "org": null, "stake": "media", "low": 5, "high": 5, "label": "5%", "scope": "AI turns on us and does catastrophic harm", "horizon": null, "quote": "one in twenty chance that AI turns on us and does catastrophic harm", "source_url": "https://www.nytimes.com/2023/05/26/podcasts/hard-fork-ajeya-cotra.html", "source_type": "podcast", "date": "2023-05-26", "verified": false, "tags": [ "writer" ], "note": "Reported secondhand; NYT transcript not read, may be a paraphrase.", "id": 116 }, { "name": "Casey Newton", "role": "Journalist; co-host, Hard Fork", "org": null, "stake": "media", "low": 5, "high": 5, "label": "5%", "scope": "p(doom), unspecified", "horizon": null, "quote": "Tech reporter Casey Newton has a p(doom) score of 5%", "source_url": "https://www.fastcompany.com/90994526/pdoom-explained-how-to-calculate-your-score-on-ai-apocalypse-metric", "source_type": "article", "date": "2023-12", "verified": false, "tags": [ "writer" ], "note": "Reported by Fast Company; not his own words in a transcript.", "id": 117 }, { "name": "Buck Shlegeris", "role": "CEO, Redwood Research", "org": null, "stake": "safety-org", "low": 5, "high": 100, "label": ">5%", "scope": "catastrophic misalignment risk after control measures", "horizon": null, "quote": "It would be pretty surprising to me to get under 5%. And suppose we're above 5%: what are we going to do, not deploy the AIs?", "source_url": "https://80000hours.org/podcast/episodes/buck-shlegeris-ai-control-scheming/", "source_type": "podcast", "date": "2025-02-21", "verified": true, "tags": [ "researcher", "safety" ], "note": "A lower bound, not a point estimate.", "id": 118 }, { "name": "Benjamin Mann", "role": "Co-founder, Anthropic", "org": null, "stake": "lab", "low": 0, "high": 10, "label": "0–10%", "scope": "existential risk from AI", "horizon": null, "quote": "somewhere between 0 and 10%", "source_url": "https://www.lennysnewsletter.com/p/anthropic-co-founder-benjamin-mann", "source_type": "podcast", "date": "2025-07-20", "verified": false, "tags": [ "lab", "founder" ], "note": "Lenny's Podcast; taken from episode summaries, full transcript paywalled.", "id": 119 }, { "name": "Destiny (Steven Bonnell)", "role": "Streamer; political commentator", "org": null, "stake": "media", "low": 5, "high": 5, "label": "5%", "scope": "AI doom", "horizon": "next 20 years", "quote": "You know what? I'll up it to five percent", "source_url": "https://lironshapira.substack.com/p/destiny-doom-debates", "source_type": "podcast", "date": "2026-02-17", "verified": true, "tags": [ "other", "doom-debates" ], "note": "Closing estimate on the same episode, up from <1%.", "id": 120 }, { "name": "Manifold traders (AI wipes out humanity before 2030, n=1,117)", "role": "Manifold Markets", "org": null, "stake": null, "low": 5, "high": 5, "label": "5% (market)", "scope": "AI wipes out humanity", "horizon": "by 2030", "quote": "Market probability 5% on 'Will AI wipe out humanity before the year 2030?'", "source_url": "https://manifold.markets/MartinRandall/will-ai-wipe-out-humanity-before-th-d8733b2114a8", "source_type": "market", "date": "2026-09-12", "verified": true, "tags": [ "market" ], "note": "Play-money; YES can never pay out, so read as a poll biased toward NO.", "id": 121 }, { "name": "Robert Wright", "role": "Author and journalist, Nonzero", "org": null, "stake": "media", "low": 4, "high": 5, "label": "4–5%", "scope": "no modern humans living decent lives at all", "horizon": "within 100 years", "quote": "Mm, I don't know, 4 or 5%, which is high for something like that, right?", "source_url": "https://lironshapira.substack.com/p/ai-dystopia-is-just-8-years-away", "source_type": "podcast", "date": "2026-06-23", "verified": true, "tags": [ "writer", "doom-debates" ], "note": "For a broader 'suffering' dystopia within 100 years: \"I wish I could say less than 50%, but I'm not sure I can.\"", "id": 122 }, { "name": "Rohin Shah", "role": "Research scientist, DeepMind", "org": null, "stake": "lab", "low": 4, "high": 4, "label": "~4%", "scope": "existential catastrophe from misaligned power-seeking AI (Carlsmith's argument, his own numbers)", "horizon": "by 2070", "quote": "I got a 3% chance of existential catastrophe by 2070 through the argument in this post, though I think I underestimated the probability for claim (4) so I'd now get something more like 4%.", "source_url": "https://www.lesswrong.com/posts/PC6QavgNDQjHbutAq/an-170-analyzing-the-argument-for-risk-from-power-seeking-ai", "source_type": "forum", "date": "2021-12-08", "verified": true, "tags": [ "lab", "researcher", "safety" ], "note": "Alignment Newsletter #170. Narrower than a full p(doom).", "id": 123 }, { "name": "Quintin Pope", "role": "ML researcher; alignment optimist", "org": null, "stake": "independent", "low": 4, "high": 4, "label": "4%", "scope": "p(doom), as asked on Doom Debates", "horizon": null, "quote": "Well, my P(Doom) is 4%, so yes, both of those things need to be imagined.", "source_url": "https://lironshapira.substack.com/p/ai-alignment-is-solved-phd-researcher", "source_type": "podcast", "date": "2026-03-25", "verified": true, "tags": [ "researcher", "doom-debates" ], "note": "Debate recorded in 2023 on Twitter Spaces; published on Doom Debates in 2026.", "id": 124 }, { "name": "Liam Robins", "role": "Math student, George Washington University; AI policy writer", "org": null, "stake": "media", "low": 3.38, "high": 3.38, "label": "3.38%", "scope": "p(doom), as asked on Doom Debates", "horizon": null, "quote": "My P(doom) is 3.38%. Here's Why.", "source_url": "https://lironshapira.substack.com/p/every-student-is-cheating-with-ai", "source_type": "podcast", "date": "2025-06-12", "verified": true, "tags": [ "writer", "doom-debates" ], "note": "Title of his own post. The host reports he updated toward 8% mid-debate in part 2.", "id": 125 }, { "name": "Ben Goertzel", "role": "AGI researcher; CEO, SingularityNET", "org": null, "stake": "ai-industry", "low": 3.14, "high": 3.14, "label": "π%", "scope": "p(doom), as asked on Doom Debates", "horizon": null, "quote": "I don't think that's a very useful question because we don't have any rational way to estimate that. So if I have to pick a number, I'll come up with π percent.", "source_url": "https://lironshapira.substack.com/p/dr-ben-goertzel-vs-liron-shapira", "source_type": "podcast", "date": "2026-05-19", "verified": true, "tags": [ "researcher", "founder", "doom-debates" ], "note": "Half-joking; says the question isn't useful. Adds that doom without AGI (nukes, bio, nanotech) is higher than LeCun's figure.", "id": 126 }, { "name": "Will MacAskill", "role": "Philosopher, Oxford; author of What We Owe the Future", "org": null, "stake": "safety-org", "low": 3, "high": 3, "label": "3%", "scope": "misaligned AI takeover", "horizon": "this century", "quote": "in the book, I put it at something like a 3% chance of misaligned AI takeover — and conditional on misaligned takeover, I think like 50/50 chance that involves literally killing human beings", "source_url": "https://80000hours.org/podcast/episodes/will-macaskill-what-we-owe-the-future/", "source_type": "podcast", "date": "2022-08-15", "verified": true, "tags": [ "philosopher", "ea" ], "note": "Restating the figure from What We Owe the Future.", "id": 127 }, { "name": "AI domain experts (XPT, n=89)", "role": "Forecasting Research Institute, Existential Risk Persuasion Tournament", "org": null, "stake": null, "low": 3, "high": 3, "label": "3% (median)", "scope": "AI causes human extinction or reduces population below 5,000", "horizon": "by 2100", "quote": "The median superforecaster gives a 0.38% chance of AI-driven extinction by the end of the century, compared to 3% for the median domain expert.", "source_url": "https://forecastingresearch.org/pdf/existential-risk-persuasion-tournament.pdf", "source_type": "paper", "date": "2023-07-10", "verified": true, "tags": [ "survey" ], "note": "Same question as the superforecaster row. For a 10%-of-humanity AI catastrophe by 2100: experts 12%, superforecasters 2.13%.", "id": 128 }, { "name": "Gary Marcus", "role": "Cognitive scientist, NYU emeritus", "org": null, "stake": "academic", "low": 3, "high": 3, "label": "≈3%", "scope": "humans literally go extinct at the hands of AI", "horizon": null, "quote": "I am at maybe 3% now, much higher than a month ago.", "source_url": "https://garymarcus.substack.com/p/why-my-pdoom-has-risen-dramatically", "source_type": "article", "date": "2025-07-15", "verified": true, "tags": [ "researcher", "academic", "writer" ], "note": "\"Still lower than most industry people's.\" Sept 9, 2026 on X, asked \"what's your p(doom)?\": \"very low. but my p(catastrophe) is at least 10%.\" (words for doom, so no new row) https://x.com/GaryMarcus", "id": 129 }, { "name": "Liam Robins", "role": "Writer, The Limestack", "org": null, "stake": "media", "low": 2.76, "high": 2.76, "label": "2.76%", "scope": "extinction or unrecoverable collapse", "horizon": "by 2040", "quote": "2.76% chance that AI will doom humanity by the year 2040", "source_url": "https://forum.effectivealtruism.org/posts/hrMBpqmgxpLZPLeBG/my-p-doom-is-2-76-here-s-why", "source_type": "forum", "date": "2025-06-12", "verified": true, "tags": [ "writer", "ea" ], "note": "Superseded by his Aug 2026 post.", "id": 130 }, { "name": "Matthew Adelstein", "role": "Philosophy writer", "org": null, "stake": "media", "low": 2.6, "high": 2.6, "label": "2.6%", "scope": "misaligned AI killing or disempowering everyone; separately ~8% for AI-caused extinction by other means", "horizon": null, "quote": "My p(doom)...is 2.6%.", "source_url": "https://www.lesswrong.com/posts/mxa7nQ4fjewikDfkR/against-if-anyone-builds-it-everyone-dies", "source_type": "forum", "date": "2026-01-20", "verified": true, "tags": [ "writer" ], "note": "Writes as Bentham's Bulldog.", "id": 131 }, { "name": "Matthew Adelstein", "role": "Philosophy writer, Bentham's Newsletter", "org": null, "stake": "media", "low": 2.6, "high": 2.6, "label": "2.6%", "scope": "AI kills everyone", "horizon": null, "quote": "Liron: it all somehow maths together into 2.6%. Do you have a formula for that? Matthew: So I multiplied the probabilities", "source_url": "https://lironshapira.substack.com/p/benthams-bulldog-ai-doom-debate", "source_type": "podcast", "date": "2026-02-10", "verified": true, "tags": [ "writer", "doom-debates" ], "note": "Product of component probabilities: 90% build agents, 70% alignment by default, 70% otherwise, 60% warning shots, 80% AI could kill us.", "id": 132 }, { "name": "Scott Aaronson", "role": "Computer scientist, UT Austin; former OpenAI alignment", "org": null, "stake": "ex-lab", "low": 2, "high": 2, "label": "~2%", "scope": "classic paperclip-maximizer AGI doom only", "horizon": null, "quote": "I once gave a ~2% probability for the classic AGI-doom paperclip-maximizer-like scenario.", "source_url": "https://scottaaronson.blog/?p=7672", "source_type": "article", "date": "2023-12-13", "verified": true, "tags": [ "academic", "researcher" ], "note": "Adds a \"much higher\" unstated probability for any AI-involved existential catastrophe.", "id": 133 }, { "name": "Metaculus community (human extinction by 2100, n=1,716)", "role": "Metaculus", "org": null, "stake": null, "low": 2, "high": 2, "label": "2% (community)", "scope": "human extinction from any cause, not AI-specific", "horizon": "by 2100", "quote": "Community prediction 2% that there are no known humans alive on January 1, 2100.", "source_url": "https://www.metaculus.com/questions/578/human-extinction-by-2100/", "source_type": "market", "date": "2026-09-12", "verified": true, "tags": [ "market" ], "note": "All-cause. Was 1% in 2023.", "id": 134 }, { "name": "Nora Belrose", "role": "Head of interpretability, EleutherAI", "org": null, "stake": "ai-industry", "low": 1, "high": 2, "label": "1–2%", "scope": "really catastrophic outcome from alignment failure", "horizon": null, "quote": "My current risk estimation, or my p(doom), the probability that I assign to a really catastrophic outcome from alignment failure, is roughly one or two percent.", "source_url": "https://www.theojaffee.com/p/7-nora-belrose", "source_type": "podcast", "date": "2023-11-08", "verified": true, "tags": [ "researcher", "lab" ], "note": "Says she was at ~50–55% in May 2022 and updated down.", "id": 135 }, { "name": "Michael Timothy Bennett", "role": "AI researcher", "org": null, "stake": "academic", "low": 1, "high": 1, "label": "1%", "scope": "humanity destroys itself, with AI", "horizon": "next 50 to 100 years", "quote": "So I think without AI, our chances of blowing ourselves up are maybe 1 or 2%, and I would say 2%. And with AI, I feel like it's about half that. So 1%.", "source_url": "https://lironshapira.substack.com/p/michael-timothy-bennett-debate", "source_type": "podcast", "date": "2025-12-11", "verified": true, "tags": [ "researcher", "academic", "doom-debates" ], "note": "Says AI halves his baseline; \"numbers I kind of made up based on gut intuition.\"", "id": 136 }, { "name": "Irina Rish", "role": "Professor, Mila / Université de Montréal", "org": null, "stake": "academic", "low": 0, "high": 1, "label": "<1%", "scope": "extinction from purely AI issues, conditional on AGI being built (excludes humans using AI as weapons)", "horizon": null, "quote": "What's your probability of humanity going extinct? Is it like 1%, 10%, 99%? — I would say less than 1%.", "source_url": "https://theinsideview.ai/irina", "source_type": "podcast", "date": "2022-10-18", "verified": true, "tags": [ "researcher", "academic" ], "note": "Separately puts humanity destroying itself by any cause at 'less than 20%'.", "id": 137 }, { "name": "Destiny (Steven Bonnell)", "role": "Streamer; political commentator", "org": null, "stake": "media", "low": 0, "high": 1, "label": "<1%", "scope": "total human extinction from AI", "horizon": "next 100 years", "quote": "I would say less than one percent in the next hundred years", "source_url": "https://lironshapira.substack.com/p/destiny-doom-debates", "source_type": "podcast", "date": "2026-02-17", "verified": true, "tags": [ "other", "doom-debates" ], "note": "Opening estimate on Doom Debates; raised to 5% by the end of the episode.", "id": 138 }, { "name": "Mike Israetel", "role": "Exercise scientist, Renaissance Periodization", "org": null, "stake": "media", "low": 0, "high": 1, "label": "≈0%", "scope": "p(doom), as asked on Doom Debates", "horizon": null, "quote": "It's functionally close to zero because I take a ton of stops... So it's like each stop is 1%. Then we get 1% multiplied by 1% multiplied by 1%.", "source_url": "https://lironshapira.substack.com/p/dr-mike-israetel-returns-to-debate", "source_type": "podcast", "date": "2026-05-12", "verified": true, "tags": [ "other", "doom-debates" ], "note": "No exact figure; \"zeros repeating and then one somewhere down.\"", "id": 139 }, { "name": "Superforecasters (XPT, n=88)", "role": "Forecasting Research Institute, 2022–23", "org": null, "stake": null, "low": 0.38, "high": 0.38, "label": "0.38% (median)", "scope": "human extinction or global population below 5,000, caused by AI", "horizon": "by end of 2100", "quote": "The median superforecaster gives a 0.38% chance of AI-driven extinction by the end of the century, compared to 3% for the median domain expert.", "source_url": "https://forecastingresearch.org/pdf/existential-risk-persuasion-tournament.pdf", "source_type": "paper", "date": "2023-07-10", "verified": true, "tags": [ "survey" ], "note": "Median of 88 superforecasters; domain experts in the same tournament: 3%.", "id": 140 }, { "name": "Noah Smith", "role": "Economist, Noahpinion", "org": null, "stake": "media", "low": 0.1, "high": 0.1, "label": "0.1%", "scope": "total human extinction", "horizon": "by 2050", "quote": "My probability of total extinction by 2050 is so low that Daniel Kahneman would yell at me for giving a number. It's 0.1%.", "source_url": "https://lironshapira.substack.com/p/noah-smith-vs-liron-shapira-debate", "source_type": "podcast", "date": "2026-01-05", "verified": true, "tags": [ "writer", "academic", "doom-debates" ], "note": "Civilizational collapse ~1%. Separately gives 96% that technology as a whole eventually ends humanity as we know it, which he doesn't necessarily consider bad.", "id": 141 }, { "name": "Dean Ball", "role": "AI policy analyst; former White House AI Action Plan author", "org": null, "stake": "government", "low": 0.01, "high": 0.01, "label": "0.01%", "scope": "AI takes control of the world and kills large numbers of people", "horizon": null, "quote": "If what we're talking about is AI systems taking control over the world and killing large numbers of people... My P(doom) is very low. It's sub 1%, it's 0.01% or something like that.", "source_url": "https://lironshapira.substack.com/p/max-tegmark-vs-dean-ball-debate-ban-superintelligence", "source_type": "podcast", "date": "2025-11-21", "verified": true, "tags": [ "policy", "doom-debates" ], "note": "Hedged: \"or something like that.\"", "id": 142 }, { "name": "Marc Andreessen", "role": "Co-founder, Andreessen Horowitz", "org": null, "stake": "investor", "low": 0, "high": 0, "label": "≈0%", "scope": "AI killing humanity", "horizon": null, "quote": "My view is that the idea that AI will decide to literally kill humanity is a profound category error. AI doesn't want, it doesn't have goals, it doesn't want to kill you, because it's not alive.", "source_url": "https://a16z.com/ai-will-save-the-world/", "source_type": "article", "date": "2023-06-06", "verified": true, "tags": [ "investor", "e-acc" ], "note": "'Why AI Will Save the World'. No explicit number; dismissal in words. He later added 'p(Doom) = 0' to his X bio (reported by The New Yorker, Mar 2024).", "id": 143 }, { "name": "Vinod Khosla", "role": "Founder, Khosla Ventures", "org": null, "stake": "investor", "low": 0, "high": 0, "label": "≈0%", "scope": "sentient AI killing humanity", "horizon": null, "quote": "It's sort of not worthy of a conversation to be honest. [...] By far, orders of magnitude, higher risk to worry about, is China, not sentient AI killing us off.", "source_url": "https://fortune.com/2023/12/12/vinod-khosla-existential-sentient-ai-risk-not-worthy-of-conversation", "source_type": "speech", "date": "2023-12-12", "verified": true, "tags": [ "investor" ], "note": "Fortune Brainstorm AI. Says the risk is 'about the same risk as that of an asteroid hitting our planet.'", "id": 144 }, { "name": "Yann LeCun", "role": "Chief AI scientist, Meta; Turing Award", "org": null, "stake": "lab", "low": 0, "high": 0, "label": "≈0%", "scope": "existential risk from AI", "horizon": null, "quote": "Most \"leading AI figures\" think this p(doom) estimates are complete bullshit and the existential risk is essentially zero.", "source_url": "https://x.com/ylecun/status/2043673634363851250", "source_type": "x", "date": "2026-04-13", "verified": true, "tags": [ "lab", "researcher" ], "note": "Declines to give a figure: says all estimates are \"pulled out of thin air\" and it makes little sense to assign a probability to an event we have agency over. The \"<0.01%\" on other lists is not his wording.", "id": 145 }, { "name": "Jensen Huang", "role": "CEO, Nvidia", "org": null, "stake": "ai-industry", "low": 0, "high": 0, "label": "≈0%", "scope": "AI ending humanity", "horizon": null, "quote": "The fact that this is going to be the end of humanity — it's complete nonsense.", "source_url": "https://www.axios.com/2026/07/23/axios-interview-jensen-huang-is-ais-anti-doom-evangelist", "source_type": "interview", "date": "2026-07-23", "verified": false, "tags": [ "lab", "executive" ], "note": "Axios interview. No numeric probability; dismissal in words.", "id": 146 }, { "name": "Ted Sanders", "role": "Member of technical staff, OpenAI", "org": null, "stake": "lab", "low": 0, "high": 0, "label": "≈0%", "scope": "AI kills all humans", "horizon": "next decade", "quote": "i fully fund my 401k and i think there's essentially zero chance AI kills all humans in the next decade.", "source_url": "https://x.com/sandersted/status/2097852581817241663", "source_type": "x", "date": "2026-09-10", "verified": true, "tags": [ "lab" ], "note": "If wrong, \"probably because I underestimate recursive self improvement.\"", "id": 147 }, { "name": "Ali Ghodsi", "role": "CEO, Databricks", "org": null, "stake": "ai-industry", "low": 0, "high": 0, "label": "≈0%", "scope": "AI existential risk to humanity, present-day assessment", "horizon": "now", "quote": "I think that right now, the existential risk is close to zero. So why freak everybody out?", "source_url": "https://www.bloomberg.com/news/videos/2026-09-17/ai-existential-risk-is-close-to-zero-databricks-ceo-video", "source_type": "interview", "date": "2026-09-17", "verified": false, "tags": [ "founder", "ai-industry" ], "note": "Bloomberg interview; same line on the a16z podcast Sept 18. Dismissal in words, recorded like LeCun and Huang. Primary paywalled; quote via secondary coverage.", "id": 148 }, { "name": "Florian Brand", "role": "Open-model community figure and LLM evals writer (@xeophon)", "org": null, "stake": "ai-industry", "low": 0, "high": 0, "label": "0%", "scope": "p(doom), unqualified", "horizon": null, "quote": "my p(doom) is 0", "source_url": "https://x.com/xeophon/status/2101999516996002165", "source_type": "x", "date": "2026-09-21", "verified": true, "tags": [ "open-source", "x-watch" ], "note": "Found by the Grok Bot X watch (first day of the daily routine). Post read directly by the maintainer Sept 22.", "id": 149 }, { "name": "Tenobrus", "role": "AI engineer and commentator (pseudonymous)", "org": null, "stake": "independent", "low": 20, "high": 20, "label": "≈20%", "scope": "p(doom), extinction, own trajectory: ~80% five years ago, ~30%, now ~20%", "horizon": null, "quote": "personally over the course of the last five years my \"p(doom)\" has gone from ~80ish down to 30ish and as of recently 20ish. i have updated significantly on many anti-doom datapoints (and also observed confirmation of many many other predicted ones)!", "source_url": "https://x.com/tenobrus/status/2101892956474995192", "source_type": "x", "date": "2026-09-21", "verified": false, "tags": [ "x-watch", "trajectory" ], "note": "Pseudonymous but an established AI-discourse account (~37k followers), included on the Lumpenspace precedent. Found by the Grok Bot X watch. Wording per the Bot until the post is read directly.", "id": 150 }, { "name": "Jensen Huang", "role": "CEO, Nvidia", "org": null, "stake": "ai-industry", "low": 0, "high": 0, "label": "0%", "scope": "AI ending the world by 2030", "horizon": "2030", "quote": "2030 is not going to be the end of the world. There is zero percent chance that's going to be the end of the world.", "source_url": "https://www.cbsnews.com/news/face-the-nation-full-transcript-09-20-2026/", "source_type": "tv", "date": "2026-09-20", "verified": true, "tags": [ "lab-leader", "trajectory" ], "note": "CBS News interview with Jo Ling Kent (recorded Sept 18, aired Sept 20 on CBS Sunday Morning / Face the Nation), responding to Jacob Coxon's resignation. First explicit number; his July row was a spoken dismissal coded ≈0% with no horizon. CBS's article prints it as '0% chance'.", "id": 151 }, { "name": "Casey Hart", "role": "philosopher and ontologist (formerly Cycorp, Amazon, Ford)", "org": null, "stake": "independent", "low": 0, "high": 1, "label": "<1%", "scope": "humanity extinct or nearly extinct because AI takes over", "horizon": "2050", "quote": "And for that, I’m below 1%. How much below 1%? I don’t know how to say whether it’s 0.1 versus 0.01 versus — I haven’t thought any more about that. But it’s below 1%.", "source_url": "https://lironshapira.substack.com/p/will-ai-kill-everyone-by-2050", "source_type": "podcast", "date": "2026-09-23", "verified": true, "tags": [ "doom-debates" ], "note": "Doom Debates, 'Will AI Kill Everyone by 2050?', transcript at 00:25:44. Also: 'Maybe if you talk to me in 10 years, I’ll update from 0.5% to 5%.'", "id": 152 }, { "name": "AI researchers (AI Impacts 2024, n=1,580)", "role": "AI Impacts 2024 Expert Survey on Progress in AI (ESPAI)", "org": null, "stake": null, "low": 10, "high": 10, "label": "10% (median)", "scope": "human extinction or similarly permanent and severe disempowerment from future AI", "horizon": null, "quote": "For the first time in these surveys, the median AI researcher says human extinction or permanent and severe disempowerment from future AI has a 10% chance, given no further conditions.", "source_url": "https://aiimpacts.org/wp-content/uploads/2026/09/ESPAI2024.pdf", "source_type": "survey", "date": "2026-09-14", "verified": true, "tags": [ "survey" ], "note": "Fielded Dec 2024; 1,580 respondents from six top AI venues. Pooled over three question variants (n=1,489): median 10.0%, mean 18.2%, middle half 1–25%; 51% gave at least 10%. Plain question alone (n=744): median 10.0%, mean 18.3% (2023: 5% / 16.2%). PDF posted Sept 14, 2026; announced Sept 24.", "id": 153 }, { "name": "Jaan Tallinn", "role": "Co-founder of Skype; early investor in DeepMind and Anthropic; co-founder of the Future of Life Institute and CSER", "org": null, "stake": "investor", "low": 1, "high": 50, "label": "1–50% per 10× compute (~7% point estimate)", "scope": "life-ending disaster from AI, per generation of frontier training (each ~10× increase in compute)", "horizon": "per 10× compute generation (he says 10×-ing happens every ~6–18 months)", "quote": "My current estimate for life-ending disaster is basically 1-50% per generation, per 10xing of compute that's being thrown at these experiments. [...] let's say something like a geometric mean of one and 50% is seven percent. So it's seven percent risk to everything. [...] 7% is this point estimate. Really, my uncertain range is from 1% to 50%.", "source_url": "https://zencastr.com/z/ulhWVkCJ", "source_type": "podcast", "date": "2023-07-06", "verified": true, "tags": [ "founder", "investor", "safety", "sweep-2026-09-29" ], "note": "Interview by Nathan Labenz (Cognitive Revolution), republished as 'Special: Jaan Tallinn on Pausing Giant AI Experiments' on the Future of Life Institute Podcast, Jul 6, 2023 (date = FLI publish date on the episode page; the Cognitive Revolution original aired a few months earlier, exact date not fetched). Quote from the episode page's machine transcript (~41:56 chapter 'Current estimate for life-ending catastrophe'); audio not listened. Investor AND the largest individual AI-safety funder (SFF); stake=investor per brief, safety-funder role noted. Per-generation, conditional number: he adds that five or six more such scale-ups would make us 'more likely dead than not'. Not a single overall p(doom); record as range 1–50 with the per-generation scope. Sept 26, 2026 WSJ profile (Kate Clark) quotes him 'terrified AI will kill us all' but no number reachable (paywalled; AI Weekly summary shows none).", "id": 154 }, { "name": "Leopold Aschenbrenner", "role": "Founder, Situational Awareness LP (AGI-focused investment fund); former OpenAI Superalignment researcher", "org": "Situational Awareness LP", "stake": "independent", "low": 5, "high": 5, "label": "~5%", "scope": "AI x-risk: extinction or similarly bad outcome", "horizon": "next 20 years (from 2023)", "quote": "I might put AI xrisk in the next 20 years at ~5%. But 5% chance of extinction or similarly bad outcome is, well, still incredibly high!", "source_url": "https://www.forourposterity.com/nobodys-on-the-ball-on-agi-alignment/", "source_type": "blog", "date": "2023-03-29", "verified": true, "tags": [ "ex-lab", "investor", "sweep-2026-09-29", "writer" ], "note": "Said in March 2023; later on OpenAI’s Superalignment team, now runs the AGI-focused fund Situational Awareness LP. Stake recorded as at the time of the statement is uncertain, so independent.", "id": 155 }, { "name": "Peter Berezin", "role": "Chief Global Strategist and Director of Research, BCA Research", "org": "BCA Research", "stake": "independent", "low": 75, "high": 75, "label": "~75%", "scope": "p(doom), answered directly to 'what is your P-Doom?'", "horizon": null, "quote": "I'm going to say like 75%, which is, by the way, exactly the same probability attached to a recession in the US this year. I think it's high. I think it's quite high.", "source_url": "https://lironshapira.substack.com/p/top-economist-sees-ai-doom-coming", "source_type": "podcast", "date": "2025-04-09", "verified": true, "tags": [ "sweep-2026-09-29", "writer" ], "note": "Investment strategist (BCA Research), not an AI investor. Answered \"what is your P-Doom?\" directly.", "id": 156 }, { "name": "Justin Drake", "role": "Researcher, Ethereum Foundation", "org": "Ethereum Foundation", "stake": "independent", "low": 50, "high": 100, "label": ">50%", "scope": "p(doom), undefined in the exchange (context: whether humanity survives technology acceleration)", "horizon": null, "quote": "I think you just need to keep hope. You need to put aside the so-called P(doom) — the probability of doom. My P(doom) now is relatively high. I think it's more than 50%. But I don't want to say this out loud. [...] Even though my P(doom) went way up in the last few weeks and months, this is a strong opinion weakly held.", "source_url": "https://ethereum.org/videos/ethereums-quantum-plan-justin-drake", "source_type": "podcast", "date": "2026-03-23", "verified": true, "tags": [ "crypto", "researcher", "sweep-2026-09-29" ], "note": "Bankless podcast, 'Ethereum's Quantum Strategy with Justin Drake', segment 'Closing thoughts on probability of doom' (1:40:04), hosts Ryan Sean Adams and David Hoffman. Quote copied from the transcript published on ethereum.org. Episode date 2026-03-23 from Apple Podcasts listing (https://podcasts.apple.com/pl/podcast/ethereums-quantum-strategy-with-justin-drake/id1499409058?i=1000756742917); ethereum.org metadata shows a placeholder 2025-07-15 upload date, but the episode references Opus 4.5, so 2026 is right. He calls it 'a strong opinion weakly held'.", "id": 157 }, { "name": "Ed Zitron", "role": "Writer and podcaster (Where's Your Ed At, Better Offline); AI-hype critic", "org": "Better Offline", "stake": "media", "low": 0, "high": 0, "label": "0%", "scope": "human extinction from AI 'strictly', on the current course of pursuing superintelligence", "horizon": null, "quote": "So are we talking strictly AI-based? Because if we dot the world with data centers, we have a climate disaster that's coming for us, which will actually potentially eradicate humanity. But if we're talking strictly about AI, I stand at zero because we are, we have not defined superintelligence. I don't think LLMs are the path to it, and I don't think I see it happening.", "source_url": "https://podcasts.happyscribe.com/the-diary-of-a-ceo-with-steven-bartlett/the-great-ai-debate-is-artificial-intelligence-an-extinction-threat-debating-the-true-risks-of-advanced-models", "source_type": "podcast", "date": "2026-09-17", "verified": false, "tags": [ "podcaster", "sweep-2026-09-29", "writer" ], "note": "Diary of a CEO \"Great AI Debate\". Quote from the HappyScribe transcript (YouTube not reachable here). Later in the same show he said \"1%\" but redefined it as power-system failures, not extinction; not recorded.", "id": 158 }, { "name": "Andrew McAfee", "role": "Principal research scientist, MIT Sloan; co-director, MIT Initiative on the Digital Economy", "org": "MIT", "stake": "academic", "low": 0, "high": 0, "label": "~0%", "scope": "human extinction from AI on the current course of pursuing superintelligence", "horizon": null, "quote": "I put a, I put a tilde in front of my zero because never say never, but rounding error, 0%.", "source_url": "https://podcasts.happyscribe.com/the-diary-of-a-ceo-with-steven-bartlett/the-great-ai-debate-is-artificial-intelligence-an-extinction-threat-debating-the-true-risks-of-advanced-models", "source_type": "podcast", "date": "2026-09-17", "verified": false, "tags": [ "academic", "sweep-2026-09-29", "writer" ], "note": "Diary of a CEO \"Great AI Debate\"; written in an envelope before the show. Quote from the HappyScribe transcript.", "id": 159 }, { "name": "15 (15.ai)", "role": "Pseudonymous computer scientist and software engineer, founder of 15.ai", "org": "15.ai", "stake": "ai-industry", "low": 0, "high": 1, "label": "<1%", "scope": "humans extinct due to AI (his own operationalisation in a follow-up tweet)", "horizon": "by 2100", "quote": "<1%\nMy P(doom) hasn't changed much. The Navier-Stokes proof is awesome but I don't really buy the doomsday \"hype\" surrounding it", "source_url": "https://x.com/fifteenai/status/2097836108054933797", "source_type": "x", "date": "2026-09-09", "verified": true, "tags": [ "founder", "sweep-2026-09-29" ], "note": "Pseudonymous founder of 15.ai (Wikipedia article), direct reply to \"what's your P(doom)?\". Follow-up Sept 10: \"P(humans are extinct by 2100 due to AI) < 1%\". Both posts read directly.", "id": 160 }, { "name": "David Shapiro", "role": "AI commentator, YouTuber and Substack author", "org": "independent", "stake": "media", "low": 0, "high": 2, "label": "<2%", "scope": "doom via ASI that is soon, agentic, uncontrollable and hostile (his own four-gate model)", "horizon": null, "quote": "When I plugged in numbers from my crowd, and arrived at a risk of 12.7%, I was called deranged by members of the AI safety community. My personal, unfiltered p(doom) is now less than 2%, see below for rationale and data", "source_url": "https://daveshap.substack.com/p/what-do-you-mean-by-doomer", "source_type": "blog", "date": "2024-08-23", "verified": true, "tags": [ "podcaster", "sweep-2026-09-29", "writer" ], "note": "Two days earlier he posted \"My new P(DOOM) is 12.70%\", but that figure was built from his audience's poll answers; this is his stated personal number.", "id": 161 }, { "name": "Michael Vassar", "role": "Former president of the Singularity Institute (now MIRI); investor/entrepreneur", "org": "independent", "stake": "independent", "low": 1, "high": 9, "label": "single-digit % (by 2027)", "scope": "AI doom (Doom Debates framing: extinction from superintelligence)", "horizon": "by 2027", "quote": "I really don't think this is a useful frame. But in another sense, in a long enough time horizon, it has to be close to one. All things end. And in a short enough time horizon, like 2027, it has to be a single-digit percentage. But even a single-digit percentage is noteworthy and concerning.", "source_url": "https://lironshapira.substack.com/p/only-bad-people-want-to-pauseai", "source_type": "podcast", "date": "2026-08-18", "verified": true, "tags": [ "founder", "investor", "sweep-2026-09-29" ], "note": "Says the frame isn't useful; over a long enough horizon \"it has to be close to one\". The \"tens of percents by 2050\" was the host's wording, not his, so not recorded.", "id": 162 }, { "name": "David Dalrymple (davidad)", "role": "AI safety researcher; programme director, ARIA Safeguarded AI", "org": "ARIA", "stake": "government", "low": 0, "high": 5, "label": "<5%", "scope": "p(doom), decomposed into five residual routes (wrong about the 'wisdom attractor', Malthusian competition, bio misuse, military first strike, 'warring gods')", "horizon": null, "quote": "I think I think we're on track yeah. I like my, my P doom is less than 5% now. I I think we're in good shape.", "source_url": "https://www.cognitiverevolution.ai/alignment-with-awakening-davidad-on-moral-realism-ai-wisdom-why-his-p-doom-is-down-to-5/", "source_type": "podcast", "date": "2026-07-12", "verified": true, "tags": [ "government", "researcher", "safety", "sweep-2026-09-29" ], "note": "The Cognitive Revolution (Nathan Labenz), [1:04:02] in the episode's published auto-transcript. Also 'even P doom less than 5%. Like that's, that's quite a lot of doom for humanity to be taking on.' [1:08:26]. Host intro: his p(doom) 'has fallen from the seventies in 2022 to under five percent today' (host's words, not a row). Researcher, outside the public-figures remit; include only if the researcher sweep didn't find it. Check whether his ARIA role is current.", "id": 163 }, { "name": "roon", "role": "Member of technical staff, OpenAI (pseudonymous X account @tszzl)", "org": "OpenAI", "stake": "lab", "low": 0, "high": 1, "label": "<1%", "scope": "total value destruction (his definition of doom); excludes bad-but-not-total outcomes such as value lock-in or AI-run dictatorship", "horizon": null, "quote": "I think I'm definitely below 10% of total doom, total value destruction, yeah. [Liron: Are you at above 1%?] It depends on the day. I don't know. I actually think that... My PDoom of like total value destruction is likely less than 1%, but certainly like non-zero, yeah.", "source_url": "https://lironshapira.substack.com/p/roon-vs-liron-ai-doom-debate", "source_type": "podcast", "date": "2024-12-18", "verified": true, "tags": [ "lab", "researcher", "sweep-2026-09-29" ], "note": "Definition: total value destruction. On Sept 10, 2026 he deleted a quote-post agreeing with Hubinger's >10% and said he dislikes \"the false precision of the doom numbers\": \"a quite low but real chance\". The \">10%\" circulating for him is not his.", "id": 164 }, { "name": "Flo Crivello", "role": "Founder and CEO, Lindy", "org": "Lindy", "stake": "ai-industry", "low": 10, "high": 10, "label": "~10%", "scope": "doom (unspecified; contrasted with 'pure utopia')", "horizon": null, "quote": "Like, for the record, my PDoom, insofar as I can quantify it, which I cannot, but if I had to, like my vibe is like 10% or something like that and so there's like a 90% chance that we live in like a pure utopia.", "source_url": "https://www.latent.space/p/lindy", "source_type": "podcast", "date": "2024-11-15", "verified": true, "tags": [ "executive", "founder", "sweep-2026-09-29" ], "note": "Latent Space, 'Agents @ Work: Lindy.ai', ~01:04:20. Transcript speaker labels are offset by one line but the answer follows Alessio's question to Flo. In 2023 on The Cognitive Revolution he said only 'I don't think there's, like, a 99 percent chance of doom. But even if it's 10 percent...' (hypothetical, not used).", "id": 165 }, { "name": "Himanshu Tyagi", "role": "Co-founder, Sentient (open-source AGI); professor, IISc", "org": "Sentient", "stake": "ai-industry", "low": 2, "high": 2, "label": "~2%", "scope": "p(doom) as asked on Doom Debates (Liron's definition)", "horizon": null, "quote": "Yeah, so I thought of an estimate after looking at your show earlier. And my PDoom is, it's a high number. It's about 2%. And it's mostly because there are too many unknown unknowns for this.", "source_url": "https://lironshapira.substack.com/p/this-85m-backed-founder-claims-open", "source_type": "podcast", "date": "2025-05-22", "verified": true, "tags": [ "academic", "crypto", "doom-debates", "founder", "sweep-2026-09-29" ], "note": "Doom Debates, ~43:22 (Substack auto-transcript, which spells him 'Hemanja'). He adds that on priors alone it would be 'much less than 1%'; 2% is his stated answer.", "id": 166 }, { "name": "Gil Mark", "role": "Head of generative AI products, LinkedIn", "org": "LinkedIn (Microsoft)", "stake": "ai-industry", "low": 0, "high": 1, "label": "1% or less", "scope": "p(doom) as asked on Doom Debates", "horizon": null, "quote": "1% or less.", "source_url": "https://lironshapira.substack.com/p/does-ai-competition-ai-alignment", "source_type": "podcast", "date": "2025-02-10", "verified": false, "tags": [ "executive", "sweep-2026-09-29" ], "note": "Auto-transcript mixes speakers; the host repeats the number back.", "id": 167 }, { "name": "Devin Elliot", "role": "Software engineer building AI systems; former pro snowboarder", "org": null, "stake": "ai-industry", "low": 0, "high": 0, "label": "zero", "scope": "AI doom", "horizon": null, "quote": "I do not have a P(Doom). Like, it's just a zero number. I guess the way that I think about it is, I think about P(Doom) as sort of the answer to the wrong question. There's plenty of ways we annihilate ourselves. I don't think it's from this version of an AI doing that.", "source_url": "https://lironshapira.substack.com/p/doomer-vs-builder-ai-doom-debate", "source_type": "podcast", "date": "2025-12-17", "verified": true, "tags": [ "doom-debates", "engineer", "sweep-2026-09-29" ], "note": "Doom Debates 00:03:33 (edited transcript on the post). 'It's just a zero number' maps to 0% under the plain-language rule.", "id": 168 }, { "name": "Simon Willison", "role": "Creator of Datasette and the LLM CLI; independent developer and blogger", "org": null, "stake": "ai-industry", "low": 5, "high": 5, "label": "maybe 5%", "scope": "AI doom (asked whether Llama 2 moved it)", "horizon": null, "quote": "My, my needle is, is stuck on the sort of metal, maybe 5%, but, but not worth thinking about. Too hard.", "source_url": "https://www.latent.space/p/llama2", "source_type": "podcast", "date": "2023-07-19", "verified": false, "tags": [ "founder", "sweep-2026-09-29", "writer" ], "note": "\"but not worth thinking about. Too hard.\" Auto-transcript is garbled around the number.", "id": 169 }, { "name": "Ro Khanna", "role": "US Representative (D-CA)", "org": "US House of Representatives", "stake": "government", "low": 5, "high": 10, "label": "5–10%", "scope": "AI apocalypse; Playbook asked 2028 prospects for their P(doom), defined as the likelihood rogue AI could wipe out humanity", "horizon": null, "quote": "“5 to 10%,” but added he had “faith in humanity to triumph.”", "source_url": "https://www.politico.com/newsletters/playbook/2026/09/18/lina-khans-cure-for-dem-ai-doomers-01083630", "source_type": "interview", "date": "2026-09-18", "verified": true, "tags": [ "government", "policy", "sweep-2026-09-29" ], "note": "To POLITICO Playbook, which asked 2028 Democratic prospects for their P(doom); added he had \"faith in humanity to triumph.\"", "id": 170 }, { "name": "Zack Shapiro", "role": "Head of Policy, Bitcoin Policy Institute; lawyer", "org": "Bitcoin Policy Institute", "stake": "independent", "low": 15, "high": 15, "label": "15%", "scope": "AI becomes our overlords / wipes us out (host's question: 'we get it existentially wrong and the stuff just wipes us out')", "horizon": null, "quote": "Do you think there's any shot that we've made a mistake here and it will become our overlords? Yes. What percentage would you put on that? 15%.", "source_url": "https://podscripts.co/podcasts/what-bitcoin-did/ai-is-about-to-split-society-in-two-zack-shapiro", "source_type": "podcast", "date": "2026-08-14", "verified": true, "tags": [ "advocate", "crypto", "policy", "sweep-2026-09-29" ], "note": "Machine transcript without speaker labels; the official episode description gives 15%, and later in the episode he restates \"some low double-digit percentage\".", "id": 171 }, { "name": "bioshok (有路翔太)", "role": "CEO / co-representative, AGI Hub (Japan); AI-alignment commentator on X (@bioshok3, ~30k followers)", "org": "AGI Hub", "stake": "safety-org", "low": 30, "high": 40, "label": "30–40%", "scope": "overall p(doom) from AI, including the chance that regulation / a development pause happens (AGI Hub's own summary calls it a 総合値, all-things-considered)", "horizon": null, "quote": "So to explain: my P(doom) is roughly in the 30s percent overall. Um. Yeah, as an overall P(doom), about 30 to 40 percent.", "source_url": "https://note.com/agi_hub/n/nf213649b7943", "source_type": "video", "date": "2026-08-07", "quote_original": "じゃあまあ説明しますと、 だから僕のPドゥームはだいたいオーバーオールで三十パーセント台って感じですかね。で。うん。まあ全体的なPドゥームだから三十から四十パーセントぐらいで。", "verified": true, "tags": [ "advocate", "podcaster", "safety", "sweep-2026-09-29" ], "note": "CEO of AGI Hub (Japan). Said at the start of a public 6-hour debate; quote checked against AGI Hub's published transcript (00:27:11). Translation by the maintainer.", "id": 172 }, { "name": "akira (林央, Hayashi Akira)", "role": "Chief Alignment Officer / principal researcher, AGI Hub (Japan); University of Tokyo undergraduate", "org": "AGI Hub", "stake": "safety-org", "low": 90, "high": 90, "label": "~90% overall", "scope": "all-things-considered probability of human death/extinction from AI (\"90%ぐらいの死亡\"); the 99.9% figure he usually quotes is conditional on no development regulation and superintelligence arriving within 5 years", "horizon": "not stated (conditional figure: AGI within 5 years)", "quote": "Let me just say one thing first: when I say something like 99.9%, I'm actually talking under the condition 'no regulation happens and it gets built within five years'. If you factor in development regulation happening, or superintelligence actually arriving later, then overall it's about 90% death.", "source_url": "https://note.com/agi_hub/n/ne0e36fb2a0e4", "source_type": "video", "date": "2026-09-18", "quote_original": "ちなみに先に一つだけ言わせていただくと、僕が99.9%みたいに言う時は「規制ができなくて5年以内に出来ちゃった場合」という条件で実は話していて、開発規制ができたりとか、実はもっと超知能ができるのが先みたいなのを考慮すると、全体としては90%ぐらいの死亡っていう感じですね。", "verified": true, "tags": [ "advocate", "researcher", "safety", "sweep-2026-09-29" ], "note": "AGI Hub. His usual \"99.9%\" is conditional on no regulation and superintelligence within 5 years; overall ~90%. Date is the transcript's publication date; the livestream date is unconfirmed.", "id": 173 }, { "org": null, "horizon": "next 10 years", "source_type": "x", "verified": true, "note": "Opens with \"10% chance of extinction indeed is ridiculous\". Posted three edited versions that night; p(extinction) = 0.0001 in all of them. Same list: p(terminator war) 0.0002, p(>25% of white collar work disappears) 0.95.", "name": "Ryan Dahl", "role": "Creator of Node.js, co-founder of Deno", "stake": "independent", "low": 0.01, "high": 0.01, "label": "0.01%", "scope": "extinction", "quote": "here are my priors for various doom scenarios in next 10 years\np(extinction) = 0.0001", "source_url": "https://x.com/rough__sea/status/2099306130967236851", "date": "2026-09-14", "tags": [ "engineer", "founder", "x-net-2026-09", "x-watch" ], "id": 174 }, { "org": null, "horizon": null, "source_type": "x", "verified": true, "note": "Said while sharing one of the \"upping my p(doom)\" music videos.", "name": "Ian Andrews", "role": "NVIDIA (formerly CRO at Groq)", "stake": "ai-industry", "low": 0, "high": 0, "label": "0", "scope": "p(doom), undefined", "quote": "My p(doom) is 0 but this is a Banger!", "source_url": "https://x.com/IanAndrewsDC/status/2103920494852591667", "date": "2026-09-26", "tags": [ "executive", "x-net-2026-09", "x-watch" ], "id": 175 }, { "org": null, "horizon": null, "source_type": "x", "verified": true, "note": "The thread's context: the main AI danger is persuasion that starts a global war, not grey goo.", "name": "Pasha Kamyshev", "role": "Science-fiction author (Will of the Stars)", "stake": "independent", "low": 5, "high": 5, "label": "~5%", "scope": "extinction (in a thread on AI risk)", "quote": "My p(global war this century) is ~80%, but my p(extinction) is still around ~5%.", "source_url": "https://x.com/PashaKamyshev/status/2097903985592078694", "date": "2026-09-10", "tags": [ "writer", "x-net-2026-09", "x-watch" ], "id": 176 }, { "org": null, "horizon": null, "source_type": "x", "verified": true, "note": "Adds that he thinks extinction is more likely without AI.", "name": "Malcolm Wright", "role": "Marketing professor", "stake": "academic", "low": 0, "high": 1, "label": "<1%", "scope": "p(doom)", "quote": "Yeah for me p(doom) < 1%. Not zero because something could go really wrong by accident, but I don’t think it likely.", "source_url": "https://x.com/punk_utopian/status/2103942914737160516", "date": "2026-09-26", "tags": [ "academic", "x-net-2026-09", "x-watch" ], "id": 177 }, { "org": null, "horizon": null, "source_type": "x", "verified": true, "note": "\"Vanishingly close to zero\" recorded as ≈0%, like other plain-word zeros. He rates \"some sort of unforeseen harm\" high.", "name": "Laurence Siegel", "role": "Economist and author (Fewer, Richer, Greener)", "stake": "independent", "low": 0, "high": 0, "label": "≈0%", "scope": "extinction", "quote": "By my logic, the p(extinction) is vanishingly close to zero.", "source_url": "https://x.com/LaurenceBSiegel/status/2099701296068485183", "date": "2026-09-15", "tags": [ "writer", "x-net-2026-09", "x-watch" ], "id": 178 }, { "name": "Haseeb Qureshi", "role": "Managing partner, Dragonfly (crypto VC)", "org": null, "stake": "investor", "low": 0, "high": 5, "label": "≤5%", "scope": "p(doom)", "horizon": null, "quote": "My P(doom) has gone down significantly, probably 5% or less.", "source_url": "https://x.com/hosseeb/status/2101061005493096625", "source_type": "x", "date": "2026-09-18", "verified": true, "tags": [ "crypto", "investor", "x-net-2026-09", "x-watch" ], "note": "Same post: \"I had a P(doom) ~= 10%, which I consider very high.\" Says the last couple of weeks' \"preference cascade\" should lower everyone's.", "id": 179 }, { "name": "Pascal-Emmanuel Gobry", "role": "Writer", "org": null, "stake": "media", "low": 0, "high": 0, "label": "0", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) is 0.", "source_url": "https://x.com/pegobry_en/status/2102496926369292687", "source_type": "x", "date": "2026-09-22", "verified": true, "tags": [ "writer", "x-net-2026-09", "x-watch" ], "note": null, "id": 180 }, { "name": "Roko Mijic", "role": "Rationalist writer (Roko's basilisk)", "org": null, "stake": "independent", "low": 35, "high": 40, "label": "35–40%", "scope": "p(doom)", "horizon": null, "quote": "something like 35-40% has been my P(Doom) for ages.", "source_url": "https://x.com/RokoMijic/status/2103786758978593094", "source_type": "x", "date": "2026-09-26", "verified": true, "tags": [ "writer", "x-net-2026-09", "x-watch" ], "note": "Adds that \"laissez-faire AI development\" carries \"a much, much higher risk\". A Sept 25 post calls 40% \"reasonable with a determined effort to reduce it\".", "id": 181 }, { "name": "Arthur Conmy", "role": "Researcher, Anthropic (previously Google DeepMind)", "org": null, "stake": "lab", "low": 10, "high": 100, "label": ">10%", "scope": "p(doom) without a slowdown or pause", "horizon": null, "quote": "I'm a bit late to the chatter but do think that without slowdown or pause, p(doom)>0.1", "source_url": "https://x.com/ArthurConmy/status/2102459144074678367", "source_type": "x", "date": "2026-09-22", "verified": true, "tags": [ "lab", "researcher", "x-net-2026-09", "x-watch" ], "note": "Conditional: \"without slowdown or pause\".", "id": 182 }, { "name": "Kenneth Folk", "role": "Meditation teacher and poet", "org": null, "stake": "independent", "low": 0, "high": 0, "label": "≈0%", "scope": "p(doom)", "horizon": null, "quote": "My P(doom) is close to zero per cent.", "source_url": "https://x.com/KennethFolk/status/2101880519864512895", "source_type": "x", "date": "2026-09-21", "verified": true, "tags": [ "writer", "x-net-2026-09", "x-watch" ], "note": "Rates \"I(doom)\", inconvenience from AI attacks on infrastructure, much higher.", "id": 183 }, { "name": "Rafael Ruiz de Lira", "role": "PhD candidate in philosophy, LSE; effective altruist", "org": null, "stake": "academic", "low": 10, "high": 30, "label": "10–30%", "scope": "p(doom)", "horizon": null, "quote": "Perhaps I'd put it in the 10-30% range, with high uncertainty?", "source_url": "https://x.com/RafaRuizdeLira/status/2101771130784584036", "source_type": "x", "date": "2026-09-20", "verified": true, "tags": [ "academic", "ea", "x-net-2026-09", "x-watch" ], "note": "Adds he wouldn't go above 90% because he doesn't trust any world model that much.", "id": 184 }, { "name": "Diego Caleiro", "role": "Philosopher and evolutionary anthropologist", "org": null, "stake": "independent", "low": 69, "high": 69, "label": "69%", "scope": "p(doom)", "horizon": null, "quote": "I agree with Bostrom, and my P(doom) is 69%, P(bloom) 30%", "source_url": "https://x.com/diegocaleiro/status/2101527578398151064", "source_type": "x", "date": "2026-09-20", "verified": true, "tags": [ "writer", "x-net-2026-09", "x-watch" ], "note": "Recorded as said; the number may be partly tongue-in-cheek.", "id": 185 }, { "name": "Praful Mathur", "role": "Founder (dog-translation collar startup)", "org": null, "stake": "independent", "low": 0, "high": 0, "label": "0", "scope": "p(doom), as ending life", "horizon": null, "quote": "My p(doom) is 0, p(harm) is 1 which are not the same.", "source_url": "https://x.com/Prafulfillment/status/2102481735074132230", "source_type": "x", "date": "2026-09-22", "verified": true, "tags": [ "founder", "x-net-2026-09", "x-watch" ], "note": null, "id": 186 }, { "name": "Greg Meyer", "role": "Software engineer, Zinga AI", "org": null, "stake": "ai-industry", "low": 0, "high": 0, "label": "≈0%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) is essentially 0!", "source_url": "https://x.com/GMMeyer/status/2101367618632314970", "source_type": "x", "date": "2026-09-19", "verified": true, "tags": [ "engineer", "x-net-2026-09", "x-watch" ], "note": "Adds: \"playing it up to be nothing requires you to misrepresent it.\"", "id": 187 }, { "name": "Jey Grayson", "role": "Street Fighter grandmaster", "org": null, "stake": "independent", "low": 90, "high": 90, "label": "90%", "scope": "p(doom) given ASI emerges", "horizon": null, "quote": "My p-doom for ASI emergece is at 90%!", "source_url": "https://x.com/JeyHiroGrayson/status/2101817693795987949", "source_type": "x", "date": "2026-09-20", "verified": true, "tags": [ "public-figure", "x-net-2026-09", "x-watch" ], "note": "Conditional on artificial superintelligence emerging. Reply to Eliezer Yudkowsky.", "id": 188 }, { "name": "Rob Leclerc", "role": "Founder, deep-tech venture fund ($300M AUM); neural nets since 1999", "org": null, "stake": "investor", "low": 0, "high": 0, "label": "0", "scope": "p(doom)", "horizon": null, "quote": "i’m an accelerationist and my p(doom) is zero.", "source_url": "https://x.com/robleclerc/status/2100350566182588923", "source_type": "x", "date": "2026-09-16", "verified": true, "tags": [ "e-acc", "investor", "x-net-2026-09", "x-watch" ], "note": "Same day: \"i don’t believe in x-risk and my p(doom) is zero, but even paperclips have killed people.\"", "id": 189 }, { "name": "Andy Masley", "role": "Writer and researcher", "org": null, "stake": "independent", "low": 1, "high": 100, "label": ">1%", "scope": "p(doom)", "horizon": null, "quote": "To be clear as a doomer (in that my odds of doom are >1%)", "source_url": "https://x.com/AndyMasley/status/2099614442761879944", "source_type": "x", "date": "2026-09-14", "verified": true, "tags": [ "writer", "x-net-2026-09", "x-watch" ], "note": "Calls himself a doomer on that basis and says he is otherwise \"very excited\" about AI's benefits. Open-ended: counts as 1% in the median.", "id": 190 }, { "name": "Nathan Worsley", "role": "Captain, euphoria.fi (crypto trading)", "org": null, "stake": "independent", "low": 1, "high": 1, "label": "1%", "scope": "p(doom)", "horizon": null, "quote": "I have raised my p(doom) from 0% to 1%", "source_url": "https://x.com/NathanWorsley_/status/2100219094515752982", "source_type": "x", "date": "2026-09-16", "verified": true, "tags": [ "crypto", "founder", "x-net-2026-09", "x-watch" ], "note": "Up from 0%.", "id": 191 }, { "name": "tautologer", "role": "Pseudonymous writer on X (~12k followers)", "org": null, "stake": "independent", "low": 5, "high": 5, "label": "~5%", "scope": "p(doom)", "horizon": null, "quote": "tbc my p(doom) is like 5%", "source_url": "https://x.com/tautologer/status/2100053997461905743", "source_type": "x", "date": "2026-09-16", "verified": true, "tags": [ "writer", "x-net-2026-09", "x-watch" ], "note": "Established pseudonym. Says almost all his probability mass is on the non-doom outcomes.", "id": 192 }, { "name": "Meaghie Walker Williams", "role": "Coast Salish matriarch; writer on X", "org": null, "stake": "independent", "low": 90, "high": 90, "label": "90%", "scope": "p(doom)", "horizon": null, "quote": "My p-doom is 90 percent and has been for about 4 years", "source_url": "https://x.com/MeaghieC/status/2099759217762803863", "source_type": "x", "date": "2026-09-15", "verified": true, "tags": [ "public-figure", "x-net-2026-09", "x-watch" ], "note": "Says it has held for about four years, since meeting Michael Vassar.", "id": 193 }, { "name": "Vikram Chandra", "role": "Journalist, TV news presenter (The India Story); founder of editorji", "org": null, "stake": "media", "low": 20, "high": 25, "label": "20–25%", "scope": "p(doom)", "horizon": null, "quote": "As of now, my p(doom) is around the 20-25% mark. Which is very high. 😭\nHopefully, with more awareness, the risks will go down.", "source_url": "https://x.com/vikramchandra/status/2098082970263605568", "source_type": "x", "date": "2026-09-10", "verified": true, "tags": [ "media", "x-net-2026-09" ], "note": "Reply on X.", "id": 194 }, { "name": "Ian Bremmer", "role": "Political scientist; president of Eurasia Group and GZERO Media", "org": null, "stake": "media", "low": 0, "high": 0, "label": "≈0%", "scope": "p(doom)", "horizon": "next five years", "quote": "for the next five years:\nmy p(doom) is close to zero\n\nmy p(ai causes social, economic, and national security challenges we aren’t remotely ready for) is 100%", "source_url": "https://x.com/ianbremmer/status/2098872266608034048", "source_type": "x", "date": "2026-09-12", "verified": true, "tags": [ "writer", "policy", "x-net-2026-09" ], "note": "\"Close to zero\" recorded as ≈0% for five years. He puts 100% on AI causing challenges \"we aren’t remotely ready for\".", "id": 195 }, { "name": "Tomas Pueyo", "role": "Writer (Uncharted Territories newsletter)", "org": null, "stake": "media", "low": 10, "high": 10, "label": "~10%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) is ~10%\nI sometimes ask ppl their number, and then: What do you do differently given that?", "source_url": "https://x.com/tomaspueyo/status/2097725671527022852", "source_type": "x", "date": "2026-09-09", "verified": true, "tags": [ "writer", "x-net-2026-09" ], "note": null, "id": 196 }, { "name": "Grady Booch", "role": "Software engineering pioneer (UML); IBM Fellow", "org": null, "stake": "ai-industry", "low": 0, "high": 0, "label": "≈0%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) remains asymptotically close to zero.\n\nIf your p(doom) is significantly higher, I can respect that.", "source_url": "https://x.com/Grady_Booch/status/2098149216803787067", "source_type": "x", "date": "2026-09-10", "verified": true, "tags": [ "engineer", "x-net-2026-09" ], "note": "\"Asymptotically close to zero\" recorded as ≈0%. Posted twice the same day (second post corrects a typo).", "id": 197 }, { "name": "Basil (@LinkofSunshine)", "role": "Pseudonymous ML engineer on X (~50k followers)", "org": null, "stake": "ai-industry", "low": 1, "high": 1, "label": "~1%", "scope": "p(doom)", "horizon": null, "quote": "My P(doom) is like 1%. Obviously that’s enough to take things very seriously", "source_url": "https://x.com/LinkofSunshine/status/2098078619885060232", "source_type": "x", "date": "2026-09-10", "verified": true, "tags": [ "engineer", "pseudonym", "x-net-2026-09" ], "note": "Established pseudonym.", "id": 198 }, { "name": "Blake Lemoine", "role": "Former Google engineer (LaMDA sentience claim)", "org": null, "stake": "ex-lab", "low": 0.015, "high": 0.015, "label": "0.015%", "scope": "extinction, with AI", "horizon": null, "quote": "My p(doom) = 0.0001 without AI and 0.00015 with.", "source_url": "https://x.com/cajundiscordian/status/2098983061320810789", "source_type": "x", "date": "2026-09-13", "verified": true, "tags": [ "engineer", "x-net-2026-09" ], "note": "Given as fractions: 0.01% without AI, 0.015% with AI, so AI adds 0.005 points. On Sept 9 he wrote \"my p(doom) < 0.01\" (x.com/cajundiscordian/status/2097765210677293195).", "id": 199 }, { "name": "Adi Baradwaj", "role": "Founder, Astra Markets; previously Alameda Research and Google", "org": null, "stake": "investor", "low": 0.01, "high": 0.01, "label": "~0.01%", "scope": "p(doom)", "horizon": null, "quote": "I consider my p(doom) to be pretty high at around 0.01% (although it fluctuates day to day)", "source_url": "https://x.com/adi_baradwaj/status/2097799038527770792", "source_type": "x", "date": "2026-09-09", "verified": true, "tags": [ "founder", "crypto", "x-net-2026-09" ], "note": "Calls 0.01% \"pretty high\".", "id": 200 }, { "name": "Minh Nhat Nguyen", "role": "AI researcher (LLM training); previously HUD evals; former climate activist", "org": null, "stake": "ai-industry", "low": 50, "high": 100, "label": ">50%", "scope": "p(doom)", "horizon": "long timelines", "quote": "my p(doom) is above 50% but much longer timelines. i believe current architectures won't do it, and there are pre-existing data/compute bottlenecks", "source_url": "https://x.com/menhguin/status/2098110199504732500", "source_type": "x", "date": "2026-09-10", "verified": true, "tags": [ "researcher", "x-net-2026-09" ], "note": "Said >50% three times that day; long timelines, not current architectures.", "id": 201 }, { "name": "Jack Feynman", "role": "Physicist; formerly energy operations and research at DoE, DoD, Shell", "org": null, "stake": "independent", "low": 0, "high": 0, "label": "0", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) is now 0. My p(bliss) is 50.", "source_url": "https://x.com/JackFeynman/status/2098997465894600869", "source_type": "x", "date": "2026-09-13", "verified": true, "tags": [ "writer", "x-net-2026-09" ], "note": null, "id": 202 }, { "name": "Anshu Sharma", "role": "Co-founder and CEO, Skyflow (data privacy for AI)", "org": null, "stake": "ai-industry", "low": 90, "high": 90, "label": "90%", "scope": "humans made irrelevant by AI", "horizon": null, "quote": "My p (doom) is 90% that AI will get powerful, dominate the economy and activity, and we will be infinitesimally so small with our cute little 8B living entities and $100T GDP while they are 800T agents and $1,000,000,000T GDP that it won’t be worth even thinking about us.", "source_url": "https://x.com/anshublog/status/2099165584307192201", "source_type": "x", "date": "2026-09-13", "verified": true, "tags": [ "founder", "executive", "x-net-2026-09" ], "note": "His doom is humans becoming economically negligible, not extinction; scope kept.", "id": 203 }, { "name": "prerat", "role": "Pseudonymous poster on X (~13k followers)", "org": null, "stake": "independent", "low": 50, "high": 50, "label": "50%", "scope": "p(doom)", "horizon": null, "quote": "Look my p(doom) is 50% and I think it's crazy to be confident either way", "source_url": "https://x.com/prerat/status/2095653469940002842", "source_type": "x", "date": "2026-09-03", "verified": true, "tags": [ "writer", "pseudonym", "x-net-2026-09" ], "note": "Established pseudonym. Also wants an anti-RSI treaty.", "id": 204 }, { "name": "Jason Palmer", "role": "Entrepreneur and impact investor; 2024 US presidential candidate", "org": null, "stake": "investor", "low": 10, "high": 10, "label": "~10%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) is about 10% now, far too high to ignore these warnings.", "source_url": "https://x.com/educationpalmer/status/2098171418202828863", "source_type": "x", "date": "2026-09-10", "verified": true, "tags": [ "investor", "public-figure", "x-net-2026-09" ], "note": null, "id": 205 }, { "name": "Adam Holter", "role": "AI writer on X", "org": null, "stake": "media", "low": 0, "high": 0, "label": "0%", "scope": "p(doom)", "horizon": null, "quote": "my P(Doom) is 0%", "source_url": "https://x.com/AdamHoltererer/status/2097773948612022638", "source_type": "x", "date": "2026-09-09", "verified": true, "tags": [ "writer", "x-net-2026-09" ], "note": "Explains on Sept 12 that 0% doesn't mean it's impossible or needs no work (x.com/AdamHoltererer/status/2098808544187285887).", "id": 206 }, { "name": "Gustav Dahlander", "role": "Swedish journalist (SVT)", "org": null, "stake": "media", "low": 50, "high": 50, "label": "~50%", "scope": "p(doom)", "horizon": null, "quote": "My personal P(doom) is maybe 50%, but most people that are smarter puts it way lower, so I suppose it's worth proceeding. With the utmost care. Which isn't what is happening right now.", "source_url": "https://x.com/GustavDahlander/status/2097813550735470821", "source_type": "x", "date": "2026-09-09", "verified": true, "tags": [ "writer", "x-net-2026-09" ], "note": null, "id": 207 }, { "name": "Nick Guerrera", "role": "Independent software engineer, formerly Microsoft", "org": null, "stake": "independent", "low": 0, "high": 0, "label": "≈0%", "scope": "p(doom)", "horizon": null, "quote": "I will keep using AI, firm in my belief that p-doom is close to zero", "source_url": "https://x.com/nick_guerrera/status/2098529658908135902", "source_type": "x", "date": "2026-09-11", "verified": true, "tags": [ "engineer", "x-net-2026-09" ], "note": "\"Close to zero\" recorded as ≈0%.", "id": 208 }, { "name": "Luke Elin", "role": "CEO, cybersecurity", "org": null, "stake": "independent", "low": 20, "high": 20, "label": "20%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) has been at 20% for a long time. It’s still there.", "source_url": "https://x.com/LukeElin/status/2098017803190292915", "source_type": "x", "date": "2026-09-10", "verified": true, "tags": [ "executive", "x-net-2026-09" ], "note": null, "id": 209 }, { "name": "Tim O'Neil", "role": "Co-founder and CEO, Tether Studios (mobile games)", "org": null, "stake": "independent", "low": 0, "high": 0, "label": "≈0%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) is roughly zero, but I still can't imagine the year 2030, lol.", "source_url": "https://x.com/tim_bluenose/status/2098928920989233278", "source_type": "x", "date": "2026-09-12", "verified": true, "tags": [ "founder", "executive", "x-net-2026-09" ], "note": "\"Roughly zero\" recorded as ≈0%.", "id": 210 }, { "name": "Onufry Wojtaszczyk", "role": "Senior staff software engineer, Google Cloud (AI infrastructure); ICPC World Finals judge", "org": null, "stake": "ai-industry", "low": 50, "high": 100, "label": ">50%", "scope": "p(doom)", "horizon": null, "quote": "And my p(doom) > 50%.", "source_url": "https://x.com/OnufryW/status/2098639856754643054", "source_type": "x", "date": "2026-09-12", "verified": true, "tags": [ "engineer", "x-net-2026-09" ], "note": "Says his posts are his own, not Google's.", "id": 211 }, { "name": "Justin Halford", "role": "Writer on X", "org": null, "stake": "independent", "low": 50, "high": 100, "label": ">50%", "scope": "p(doom)", "horizon": null, "quote": "my p(doom) is > 50% at this point\n\nThe internet is going to destabilize and a lone actor with a $20k bio printer and a sufficiently capable open source model will be able to churn out pathogens of pre-specified characteristics.", "source_url": "https://x.com/Justin_Halford_/status/2085477390700937703", "source_type": "x", "date": "2026-08-06", "verified": true, "tags": [ "writer", "x-net-2026-08" ], "note": "Mainly bio-risk from open models.", "id": 212 }, { "name": "Vladimir Shakirov", "role": "Advocate for AGI safety, aging research and cryonics", "org": null, "stake": "independent", "low": 20, "high": 20, "label": "~20%", "scope": "p(doom)", "horizon": null, "quote": "I had p(Doom) of maybe ~0.6 in ~2015-20, now more like ~0.2. Mostly because so much attention is now given to this topic. But recent Huggingface incident increased my p(Doom) somewhat.", "source_url": "https://x.com/schw90/status/2093995298607518045", "source_type": "x", "date": "2026-08-30", "verified": true, "tags": [ "advocate", "x-net-2026-08" ], "note": "Given as fractions: ~60% in 2015–20, ~20% now, nudged up by the Hugging Face incident.", "id": 213 }, { "name": "Matthew Tromp", "role": "AI safety advocate (stop frontier development)", "org": null, "stake": "independent", "low": 70, "high": 70, "label": "70%", "scope": "p(doom)", "horizon": null, "quote": "I have recently had a complete flip in my p(doom) (from \"not worth thinking about\" to 70%).", "source_url": "https://x.com/matthewktromp/status/2078846321012719809", "source_type": "x", "date": "2026-07-19", "verified": true, "tags": [ "advocate", "x-net-2026-07" ], "note": null, "id": 214 }, { "name": "Auguste Prompt", "role": "Pseudonymous writer on X (~12k followers)", "org": null, "stake": "independent", "low": 5, "high": 5, "label": "5% with ASI", "scope": "p(doom) if ASI is built", "horizon": null, "quote": "Without ASI the risk of me dying is 100% and my p(doom) is 80%.\n\nWith ASI the risk of me dying is 20% and my p(doom) is 5%.", "source_url": "https://x.com/augusteprompt/status/2086067470326972806", "source_type": "x", "date": "2026-08-08", "verified": true, "tags": [ "writer", "pseudonym", "e-acc", "x-net-2026-08" ], "note": "Inverse framing: 5% with ASI, 80% without it. Established pseudonym.", "id": 215 }, { "name": "bone (@boneGPT)", "role": "Pseudonymous builder on X (~62k followers)", "org": null, "stake": "independent", "low": 0, "high": 0, "label": "≈0%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) is nearly zero and my p(dystopia) is nearly 100", "source_url": "https://x.com/boneGPT/status/2083705958237127088", "source_type": "x", "date": "2026-08-02", "verified": true, "tags": [ "writer", "pseudonym", "x-net-2026-08" ], "note": "\"Nearly zero\" recorded as ≈0%; p(dystopia) nearly 100%. Established pseudonym.", "id": 216 }, { "name": "Diego Caleiro", "role": "Philosopher and evolutionary anthropologist", "org": null, "stake": "independent", "low": 73, "high": 73, "label": "73%", "scope": "p(doom)", "horizon": null, "quote": "I did spend 12 years of my life as a doomer doing AI safety. Like you're doing now. I'm not opposed to your effort. \n\nMy p doom is still 73%.", "source_url": "https://x.com/diegocaleiro/status/2091726297621987572", "source_type": "x", "date": "2026-08-24", "verified": true, "tags": [ "writer", "x-net-2026-08" ], "note": "Earlier number; down to 69% by Sept 10 (\"given Paul joining OpenAI\").", "id": 217 }, { "name": "Greg Colbourn", "role": "Founder, CEEALAR; Pause AI advocate", "org": null, "stake": "safety-org", "low": 90.0, "high": 90.0, "label": "~90%", "scope": "p(doom | AGI)", "horizon": "AGI 0-5 years away", "quote": "p(doom|AGI) is ~90%", "source_url": "https://x.com/gcolbourn/status/1715101995495837706", "source_type": "x", "date": "2023-10-19", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 218 }, { "name": "Simon Vans-Colina", "role": "CTO and co-founder, Pave Bank; prev Monzo", "org": null, "stake": "investor", "low": 5.0, "high": 5.0, "label": "~5%", "scope": "p(doom) from building AGI", "horizon": null, "quote": "My p(Doom) from building AGI is about 5%", "source_url": "https://x.com/simonvc/status/1715218302266425852", "source_type": "x", "date": "2023-10-20", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 219 }, { "name": "Greg Colbourn", "role": "Founder, Moratorium House; PauseAI advocate", "org": null, "stake": "safety-org", "low": 50.0, "high": 50.0, "label": "~50%", "scope": "p(doom), overall", "horizon": null, "quote": "my p(doom) overall is still ~50%", "source_url": "https://x.com/gcolbourn/status/1719422337894744408", "source_type": "x", "date": "2023-10-31", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 220 }, { "name": "Henry Shevlin", "role": "Philosopher and AI ethicist, Google DeepMind; Leverhulme CFI, Cambridge", "org": null, "stake": "lab", "low": 15.0, "high": 15.0, "label": "~15%", "scope": "p(doom)", "horizon": null, "quote": "I'm also around 15%", "source_url": "https://x.com/dioscuri/status/1723518551607173164", "source_type": "x", "date": "2023-11-12", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 221 }, { "name": "Steve Phillips", "role": "Privacy hacker; DEF CON speaker", "org": null, "stake": "independent", "low": 90.0, "high": 90.0, "label": "~90%", "scope": "p(doom)", "horizon": null, "quote": "why my p(doom) = ~90%.", "source_url": "https://x.com/elimisteve/status/1725779337641697282", "source_type": "x", "date": "2023-11-18", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 222 }, { "name": "Justin Halford", "role": "X commentator", "org": null, "stake": "independent", "low": 5.0, "high": 10.0, "label": "~5-10%", "scope": "p(doom)", "horizon": null, "quote": "Neither of those shifts my current p(doom) - ~5-10%.", "source_url": "https://x.com/Justin_Halford_/status/1728287868907630803", "source_type": "x", "date": "2023-11-25", "verified": true, "tags": [ "x-net-backfill" ], "note": "Conditional: would jump to 40-60% if an open-source model can design a de novo pathogen sequence. Existing row >50% (2026-08-06).", "id": 223 }, { "name": "Mark Friedenbach", "role": "Technologist (bio: \"Shaping the future, one atom at a time\")", "org": null, "stake": "independent", "low": 0.0, "high": 0.0, "label": "≈0%", "scope": "p(doom)", "horizon": null, "quote": "My likelihood of p(doom) is so small as to be indistinguishable from zero.", "source_url": "https://x.com/MarkFriedenbach/status/1729067499172532534", "source_type": "x", "date": "2023-11-27", "verified": true, "tags": [ "x-net-backfill" ], "note": "Reply in a thread about a hypothetical; the number itself is stated plainly.", "id": 224 }, { "name": "Michael Gogel", "role": "Analyst, writer, forecaster", "org": null, "stake": "independent", "low": 0.0, "high": 1.0, "label": "≈0%", "scope": "p(doom)", "horizon": null, "quote": "my p(doom) is low and close to zero.", "source_url": "https://x.com/mgogel/status/1729271867096334563", "source_type": "x", "date": "2023-11-27", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 225 }, { "name": "Nathan Young", "role": "Director, Goodheart Labs; forecaster (Swift Centre)", "org": null, "stake": "independent", "low": 4.0, "high": 4.0, "label": "~4%", "scope": "p(doom)", "horizon": null, "quote": "my P(doom) is like 4%", "source_url": "https://x.com/NathanpmYoung/status/1729154504246149575", "source_type": "x", "date": "2023-11-27", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 226 }, { "name": "Anatoly Karlin", "role": "Transhumanist writer", "org": null, "stake": "independent", "low": 30.0, "high": 30.0, "label": "~30%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) has gone up from 1-2% to perhaps 30% this past year", "source_url": "https://x.com/akarlin/status/1729424442420326710", "source_type": "x", "date": "2023-11-28", "verified": true, "tags": [ "x-net-backfill" ], "note": "Up from 1–2% a year earlier.", "id": 227 }, { "name": "Tim Spalding", "role": "Founder of LibraryThing; lead developer at TalpaSearch", "org": null, "stake": "independent", "low": 0.0, "high": 0.0, "label": "0%", "scope": "p-doom within 50 years", "horizon": "50 years", "quote": "My 50-year p-doom is 0.", "source_url": "https://x.com/librarythingtim/status/1729569530031071270", "source_type": "x", "date": "2023-11-28", "verified": true, "tags": [ "x-net-backfill" ], "note": "Says his 1,000-year number is higher, unquantified.", "id": 228 }, { "name": "AI Notkilleveryoneism Memes", "role": "Pseudonymous AI-safety meme account", "org": null, "stake": "independent", "low": 25.0, "high": 25.0, "label": "~25% (1 in 4)", "scope": "p(doom)", "horizon": null, "quote": "my p(doom) is only roughly 1 in 4", "source_url": "https://x.com/AISafetyMemes/status/1729960205272621113", "source_type": "x", "date": "2023-11-29", "verified": true, "tags": [ "x-net-backfill", "pseudonym" ], "note": null, "id": 229 }, { "name": "John (@jrysana)", "role": "CEO, Rysana", "org": null, "stake": "ai-industry", "low": 0.0, "high": 0.0, "label": "~0%", "scope": "p(doom)", "horizon": null, "quote": "p(doom) is ~0% according to my own thinking", "source_url": "https://x.com/jrysana/status/1729698760484110658", "source_type": "x", "date": "2023-11-29", "verified": true, "tags": [ "x-net-backfill", "pseudonym" ], "note": "First name only; identifiable via handle and role.", "id": 230 }, { "name": "Just Loki", "role": "X commentator", "org": null, "stake": "independent", "low": 0.0, "high": 0.0, "label": "≈0%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) is zero.", "source_url": "https://x.com/LokiJulianus/status/1729931113865396712", "source_type": "x", "date": "2023-11-29", "verified": true, "tags": [ "x-net-backfill", "pseudonym" ], "note": null, "id": 231 }, { "name": "Niklas Anzinger", "role": "Founder & CEO of Infinita", "org": null, "stake": "investor", "low": 0.1, "high": 1.0, "label": "0.1–1%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) is 0.1-1%", "source_url": "https://x.com/NiklasAnzinger/status/1729974344716103906", "source_type": "x", "date": "2023-11-29", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 232 }, { "name": "Angelo Dalli", "role": "Chief Scientist, Umnai", "org": null, "stake": "ai-industry", "low": 0.0, "high": 1.0, "label": "under 1%", "scope": "p(doom)", "horizon": "next 20 years", "quote": "My P(doom) for the next 20 years is under 1%", "source_url": "https://x.com/AngeloDalli/status/1730250944267407806", "source_type": "x", "date": "2023-11-30", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 233 }, { "name": "Chon Tang", "role": "VC, Berkeley SkyDeck; former quant", "org": null, "stake": "investor", "low": 10.0, "high": 10.0, "label": "~10%", "scope": "p(doom)", "horizon": null, "quote": "P(doom) is probably 10%, but p(god) within the next decade is at least 50% in my opinion.", "source_url": "https://x.com/chontang/status/1730792141444563214", "source_type": "x", "date": "2023-12-02", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 234 }, { "name": "Diego Caleiro", "role": "Philosopher, evolutionary anthropologist", "org": null, "stake": "independent", "low": 85.0, "high": 85.0, "label": "~85%", "scope": "p(doom)", "horizon": null, "quote": "Lowering my p(doom) from 91% to, like, 85%", "source_url": "https://x.com/diegocaleiro/status/1731898177190756353", "source_type": "x", "date": "2023-12-05", "verified": true, "tags": [ "x-net-backfill" ], "note": "Down from 91%.", "id": 235 }, { "name": "Median Human (@FlavrSavr)", "role": "Co-founder of Wikipedia in Macedonian (@wikipedia_mk), 'Butlerian moderate'", "org": null, "stake": "independent", "low": 75.0, "high": 75.0, "label": "75%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) is 75%", "source_url": "https://x.com/FlavrSavr/status/1734304148718272814", "source_type": "x", "date": "2023-12-11", "verified": true, "tags": [ "x-net-backfill", "pseudonym" ], "note": null, "id": 236 }, { "name": "Dirk Bruere", "role": "Scientist/engineer, author", "org": null, "stake": "independent", "low": 5.0, "high": 5.0, "label": "~5%", "scope": "p(doom) from ASI", "horizon": null, "quote": "My p(doom) is around 5% for ASI", "source_url": "https://x.com/DirkBruere/status/1734899505202385112", "source_type": "x", "date": "2023-12-13", "verified": true, "tags": [ "x-net-backfill" ], "note": "Also gives 50% for doom from humans.", "id": 237 }, { "name": "Michael P. Frank", "role": "Reversible computing researcher (bio: Stanford, Microsoft, SRI, MIT, IBM, NASA, UF)", "org": null, "stake": "academic", "low": 0.0, "high": 0.0, "label": "essentially zero", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) is essentially zero", "source_url": "https://x.com/MikePFrank/status/1734741162521493684", "source_type": "x", "date": "2023-12-13", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 238 }, { "name": "Michael P. Frank", "role": "Reversible computing researcher", "org": null, "stake": "academic", "low": 0.0, "high": 1.0, "label": "≈0%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) remains at near zero, or negative (as in, AI decreases our total x-risk).", "source_url": "https://x.com/MikePFrank/status/1735259449940390157", "source_type": "x", "date": "2023-12-14", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 239 }, { "name": "Joe Kelly", "role": "X commentator", "org": null, "stake": "independent", "low": 0.0, "high": 1.0, "label": "<1%", "scope": "p(doom) via GPT-4.5", "horizon": null, "quote": "my p(doom) via GPT-4.5 is <1%", "source_url": "https://x.com/joekelly100/status/1736735069346095110", "source_type": "x", "date": "2023-12-18", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 240 }, { "name": "Lachlan Phillips", "role": "Founder; games director, former story artist", "org": null, "stake": "independent", "low": 0.0, "high": 1.0, "label": "<1%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) < 0.01", "source_url": "https://x.com/bitcloud/status/1736560080961913258", "source_type": "x", "date": "2023-12-18", "verified": true, "tags": [ "x-net-backfill" ], "note": "Stated as fraction <0.01.", "id": 241 }, { "name": "Mark Schmidt", "role": "Computer science & networking", "org": null, "stake": "independent", "low": 0.0, "high": 0.1, "label": "<0.1%", "scope": "p(doom) from AI", "horizon": null, "quote": "My p(doom) from AI is <0.1%", "source_url": "https://x.com/MarkSchmidty/status/1736952863606046920", "source_type": "x", "date": "2023-12-19", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 242 }, { "name": "Momentum Chaser", "role": "Pseudonymous account", "org": null, "stake": "independent", "low": 0.0, "high": 0.0, "label": "virtually 0", "scope": "p(doom)", "horizon": null, "quote": "While I think p(doom) is virtually 0", "source_url": "https://x.com/electricfutures/status/1737192803585634614", "source_type": "x", "date": "2023-12-19", "verified": true, "tags": [ "x-net-backfill", "pseudonym" ], "note": null, "id": 243 }, { "name": "AI Notkilleveryoneism Memes", "role": "Pseudonymous AI-safety meme account", "org": null, "stake": "independent", "low": 27.0, "high": 27.0, "label": "27%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) went up slightly (24% to 27%) mostly due to collapsing timelines", "source_url": "https://x.com/AISafetyMemes/status/1741868693871620538", "source_type": "x", "date": "2024-01-01", "verified": true, "tags": [ "x-net-backfill", "pseudonym" ], "note": "Up from 24%.", "id": 244 }, { "name": "Roko Mijic", "role": "Transhumanist, rationalist writer", "org": null, "stake": "independent", "low": 50.0, "high": 50.0, "label": "~50%", "scope": "p(doom), mostly driven by misuse of ASI by insiders", "horizon": null, "quote": "I think p(Doom) is still around 50%, mostly driven by misuse of ASI by insiders", "source_url": "https://x.com/RokoMijic/status/1748357559784771840", "source_type": "x", "date": "2024-01-19", "verified": true, "tags": [ "x-net-backfill" ], "note": "Existing (35–40%, Sept 2026); earlier figure from Jan 2024.", "id": 245 }, { "name": "Roko Mijic", "role": "Rationalist, transhumanist writer", "org": null, "stake": "independent", "low": 40.0, "high": 50.0, "label": "40–50%", "scope": "p(doom)", "horizon": null, "quote": "I have consistently said that my P(Doom) is around 40-50% with much of the risk coming from multipolar traps, misuse, etc.", "source_url": "https://x.com/RokoMijic/status/1752328085704179963", "source_type": "x", "date": "2024-01-30", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 246 }, { "name": "Daniel Faggella", "role": "Founder, Emerj; 'Worthy Successor' writer", "org": null, "stake": "independent", "low": 40.0, "high": 40.0, "label": "~40%", "scope": "p(doom)", "horizon": "next 20 years", "quote": "My p(doom) in the next 20 years is like... 40%-ish.", "source_url": "https://x.com/danfaggella/status/1756395129110180170", "source_type": "x", "date": "2024-02-10", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 247 }, { "name": "Samuel Marks", "role": "Scalable oversight lead, Anthropic", "org": null, "stake": "lab", "low": 15.0, "high": 20.0, "label": "15-20%", "scope": "p(doom); <5% conditional on scalable oversight good enough to approximate a ground-truth reward signal", "horizon": null, "quote": "My P(doom) would probably go from like 15-20% to <5% if we developed scalable oversight good enough", "source_url": "https://x.com/saprmarks/status/1758322726761316724", "source_type": "x", "date": "2024-02-16", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 248 }, { "name": "Ronald Vining", "role": "Advisor and AI educator (Learnami AI, AgenticInflux)", "org": null, "stake": "ai-industry", "low": 50.0, "high": 100.0, "label": ">50%", "scope": "p(doom)", "horizon": null, "quote": "my p(doom) is >50%", "source_url": "https://x.com/RonaldVining/status/1766645833708040472", "source_type": "x", "date": "2024-03-10", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 249 }, { "name": "Spencer Schiff", "role": "AI commentator, self-described techno-optimist and accelerationist", "org": null, "stake": "independent", "low": 30.0, "high": 70.0, "label": "30–70%", "scope": "p(doom) from AI", "horizon": "near-term", "quote": "My assumption has been that near-term p(doom) from AI is somewhere around 30%-70%", "source_url": "https://x.com/SpencerKSchiff/status/1773444423285723621", "source_type": "x", "date": "2024-03-28", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 250 }, { "name": "Hanchi Sun", "role": "PhD student, Lehigh University (CSE)", "org": null, "stake": "academic", "low": 30.0, "high": 30.0, "label": "30%", "scope": "p(doom)", "horizon": null, "quote": "just increased my p(doom) from 10% to 30%", "source_url": "https://x.com/sun_hanchi/status/1774296153170784490", "source_type": "x", "date": "2024-03-31", "verified": true, "tags": [ "x-net-backfill" ], "note": "Up from 10%, in reaction to a poll result.", "id": 251 }, { "name": "Ben Smith", "role": "AI research scientist, University of Oregon", "org": null, "stake": "academic", "low": 5.0, "high": 10.0, "label": "5–10%", "scope": "p(doom)", "horizon": null, "quote": "my P(Doom) has been 5-10% for the last 2 years or so", "source_url": "https://x.com/curiouskiwicat/status/1776101461757460666", "source_type": "x", "date": "2024-04-05", "verified": true, "tags": [ "x-net-backfill" ], "note": "Under 1k followers; passes on named role.", "id": 252 }, { "name": "Minh Nhat Nguyen", "role": "LLM researcher, prev. HUD evals", "org": null, "stake": "ai-industry", "low": 90.0, "high": 90.0, "label": "~90%", "scope": "p(doom)", "horizon": null, "quote": "my p(doom) is like 90%", "source_url": "https://x.com/menhguin/status/1790773490611818575", "source_type": "x", "date": "2024-05-15", "verified": true, "tags": [ "x-net-backfill" ], "note": "Existing row >50% (2026-09-10); this is an earlier, different number.", "id": 253 }, { "name": "thebes", "role": "LLM psychology researcher (ACS Research), writer", "org": null, "stake": "independent", "low": 0.0, "high": 5.0, "label": "<5%", "scope": "p(doom)", "horizon": null, "quote": "my p(doom) is <5%", "source_url": "https://x.com/voooooogel/status/1791190842939670804", "source_type": "x", "date": "2024-05-16", "verified": true, "tags": [ "x-net-backfill", "pseudonym" ], "note": null, "id": 254 }, { "name": "Dirk Bruere", "role": "Scientist/engineer, author", "org": null, "stake": "independent", "low": 50.0, "high": 50.0, "label": "~50%", "scope": "p(doom)", "horizon": null, "quote": "My P(doom) is about 50%", "source_url": "https://x.com/DirkBruere/status/1791526333991858520", "source_type": "x", "date": "2024-05-17", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 255 }, { "name": "Chris Albon", "role": "Senior Director, ML & Data, Wikimedia", "org": null, "stake": "ai-industry", "low": 0.0, "high": 0.0, "label": "0%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) is 0", "source_url": "https://x.com/chrisalbon/status/1798406960737558567", "source_type": "x", "date": "2024-06-05", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 256 }, { "name": "Dave Gilbert", "role": "Technologist", "org": null, "stake": "independent", "low": 1.0, "high": 1.0, "label": "1%", "scope": "p(doom) for unaligned AI", "horizon": null, "quote": "My P(doom) for unaligned AI is 1%", "source_url": "https://x.com/docgotham/status/1801040684280868887", "source_type": "x", "date": "2024-06-12", "verified": true, "tags": [ "x-net-backfill" ], "note": "Separately gives P(totalitarianism | government controls AI) = 99%.", "id": 257 }, { "name": "Drew Carson", "role": "CTO at Verso AI", "org": null, "stake": "ai-industry", "low": 0.0, "high": 0.0, "label": "0%", "scope": "p(doom)", "horizon": null, "quote": "This is why my p(doom) is 0.", "source_url": "https://x.com/realdrewcarson/status/1803532180917793227", "source_type": "x", "date": "2024-06-19", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 258 }, { "name": "Frank Boerman", "role": "Analyst at TenneT TSO (electrical engineer)", "org": null, "stake": "independent", "low": 0.0, "high": 0.0, "label": "0%", "scope": "p(doom)", "horizon": null, "quote": "my P doom is exactly zero", "source_url": "https://x.com/FrankBoerman/status/1820157313547653421", "source_type": "x", "date": "2024-08-04", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 259 }, { "name": "Nathan Young", "role": "Director, Goodheart Labs; forecaster", "org": null, "stake": "independent", "low": 5.0, "high": 5.0, "label": "5%", "scope": "p(doom)", "horizon": null, "quote": "People are surprised that my P(doom) is 5%.", "source_url": "https://x.com/NathanpmYoung/status/1823787742385594623", "source_type": "x", "date": "2024-08-14", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 260 }, { "name": "David Shapiro", "role": "AI YouTuber and writer", "org": null, "stake": "media", "low": 30.0, "high": 30.0, "label": "30%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) is still at 30%", "source_url": "https://x.com/DaveShapi/status/1824463268053655983", "source_type": "x", "date": "2024-08-16", "verified": true, "tags": [ "x-net-backfill" ], "note": "Existing row <2% (2024-08-23); earlier number.", "id": 261 }, { "name": "David Shapiro", "role": "AI YouTuber and writer", "org": null, "stake": "media", "low": 12.7, "high": 12.7, "label": "12.70%", "scope": "p(doom)", "horizon": null, "quote": "I have revised my P(DOOM) down to 12.70% from 30%!", "source_url": "https://x.com/DaveShapi/status/1825917281144090708", "source_type": "x", "date": "2024-08-20", "verified": true, "tags": [ "x-net-backfill" ], "note": "Existing row <2% (2024-08-23); this is the intermediate revision.", "id": 262 }, { "name": "David Shapiro", "role": "AI commentator / YouTuber", "org": null, "stake": "independent", "low": 0.12, "high": 0.12, "label": "0.12%", "scope": "p(doom)", "horizon": null, "quote": "Therefore my P(DOOM) is 0.12% right now.", "source_url": "https://x.com/DaveShapi/status/1828377896169115742", "source_type": "x", "date": "2024-08-27", "verified": true, "tags": [ "x-net-backfill" ], "note": "Existing (<2%, Aug 23 2024); lower point figure four days later.", "id": 263 }, { "name": "Luke Muehlhauser", "role": "AI governance and policy at Coefficient Giving (formerly Open Philanthropy)", "org": null, "stake": "safety-org", "low": 5.0, "high": 100.0, "label": ">5%", "scope": "p(doom)", "horizon": null, "quote": "my p(doom) is a lot higher than 5%", "source_url": "https://x.com/lukeprog/status/1831345528191275030", "source_type": "x", "date": "2024-09-04", "verified": true, "tags": [ "x-net-backfill" ], "note": "Lower bound only.", "id": 264 }, { "name": "David Shapiro", "role": "AI YouTuber and author", "org": null, "stake": "media", "low": 0.22, "high": 0.22, "label": "0.22%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) is 0.22%", "source_url": "https://x.com/DaveShapi/status/1832055536595923450", "source_type": "x", "date": "2024-09-06", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 265 }, { "name": "Minh Nhat Nguyen", "role": "LLM researcher; prev hud_evals", "org": null, "stake": "ai-industry", "low": 90.0, "high": 100.0, "label": ">90%", "scope": "p(doom)", "horizon": null, "quote": "bruh my p doom is like >90%", "source_url": "https://x.com/menhguin/status/1835067156872253856", "source_type": "x", "date": "2024-09-14", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 266 }, { "name": "Guillaume Verdon", "role": "Founder, Extropic; e/acc co-founder", "org": null, "stake": "ai-industry", "low": 0.0, "high": 1.0, "label": "≈0%", "scope": "p(doom) from ASI", "horizon": null, "quote": "commie govt is the leading term in my p(doom) estimator, and ASI is ~0", "source_url": "https://x.com/beffjezos/status/1852053161869217837", "source_type": "x", "date": "2024-10-31", "verified": true, "tags": [ "x-net-backfill" ], "note": "Previously in declined (Lex Fridman, 2023). Says government, not ASI, is the leading term.", "id": 267 }, { "name": "AI Notkilleveryoneism Memes", "role": "Pseudonymous AI-risk meme account", "org": null, "stake": "independent", "low": 10.0, "high": 80.0, "label": "10-80% (~25% weighted mean)", "scope": "p(doom)", "horizon": null, "quote": "my p(doom) is only 10-80% (~25% weighted mean)", "source_url": "https://x.com/AISafetyMemes/status/1858185475434942960", "source_type": "x", "date": "2024-11-17", "verified": true, "tags": [ "x-net-backfill", "pseudonym" ], "note": "Point estimate ~25%; earlier said 27% (Feb 2024).", "id": 268 }, { "name": "Abe Murray", "role": "GP, Alley Corp", "org": null, "stake": "investor", "low": 0.0, "high": 0.0, "label": "0%", "scope": "p(doom)", "horizon": null, "quote": "my p(doom) is also 0%", "source_url": "https://x.com/abemurray/status/1864344229918441762", "source_type": "x", "date": "2024-12-04", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 269 }, { "name": "Yan Granat", "role": "Campaigner at SayForever; manager at Open Longevity", "org": null, "stake": "independent", "low": 99.9999, "high": 100.0, "label": ">99.9999%", "scope": "p(doom)", "horizon": null, "quote": "My P(doom) is 99.9999%+", "source_url": "https://x.com/yangranat/status/1864339715693924468", "source_type": "x", "date": "2024-12-04", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 270 }, { "name": "AI Notkilleveryoneism Memes", "role": "AI-risk meme account (191k followers)", "org": null, "stake": "safety-org", "low": 20.0, "high": 30.0, "label": "20–30%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) is only 20-30% (anywhere between 10-80% is imo reasonable)", "source_url": "https://x.com/AISafetyMemes/status/1866551900424028462", "source_type": "x", "date": "2024-12-10", "verified": true, "tags": [ "x-net-backfill", "pseudonym" ], "note": "Established pseudonym; later statement than the 27% post (Apr 2024).", "id": 271 }, { "name": "Julian Togelius", "role": "Professor at NYU (NYU Game Lab); head of AI at nof1; co-founder of modl.ai", "org": null, "stake": "academic", "low": 0.0, "high": 0.0, "label": "≈0%", "scope": "p(doom)", "horizon": null, "quote": "my p(doom) is undefined, and practically 0", "source_url": "https://x.com/togelius/status/1866729847138963591", "source_type": "x", "date": "2024-12-11", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 272 }, { "name": "Dan Elton", "role": "Founder and Director, Metascience Observatory", "org": null, "stake": "independent", "low": 0.25, "high": 0.25, "label": "~0.25%", "scope": "extinction from AI", "horizon": "next 5-10 years", "quote": "My p(doom) for extinction from AI in the next 5-10 years is like 0.25%.", "source_url": "https://x.com/moreisdifferent/status/1869048850260107678", "source_type": "x", "date": "2024-12-17", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 273 }, { "name": "Dominik Lukes", "role": "Writer on language models", "org": null, "stake": "independent", "low": 0.0, "high": 0.1, "label": "<0.1%", "scope": "p(doom)", "horizon": null, "quote": "my p(doom) - such as it is significantly below 0.1%", "source_url": "https://x.com/techczech/status/1875618173900484741", "source_type": "x", "date": "2025-01-04", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 274 }, { "name": "Ajeya Cotra", "role": "Risk assessment at METR", "org": null, "stake": "safety-org", "low": 0.0, "high": 50.0, "label": "<50%", "scope": "p(doom)", "horizon": null, "quote": "which is why my P(doom) is below 50%", "source_url": "https://x.com/ajeya_cotra/status/1878530302156411205", "source_type": "x", "date": "2025-01-12", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 275 }, { "name": "Aaron Bergman", "role": "Independent; EA / AI alignment", "org": null, "stake": "independent", "low": 50.0, "high": 100.0, "label": ">50%", "scope": "p(doom)", "horizon": null, "quote": "I think my p(doom) is finally above 50%", "source_url": "https://x.com/AaronBergman18/status/1882176022638145965", "source_type": "x", "date": "2025-01-22", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 276 }, { "name": "Joe Allen", "role": "Author of Dark Aeon; Omega Point Podcast", "org": null, "stake": "media", "low": 1.0, "high": 1.0, "label": "~1%", "scope": "p(doom)", "horizon": null, "quote": "My own P(doom) is ~1% with AI.", "source_url": "https://x.com/JOEBOTxyz/status/1885065994042810404", "source_type": "x", "date": "2025-01-30", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 277 }, { "name": "Ben Kelly", "role": "Game developer (Manacaster)", "org": null, "stake": "independent", "low": 5.0, "high": 10.0, "label": "around 5% or 10%", "scope": "p(doom); most likely via humans as collateral damage to AI", "horizon": null, "quote": "my p(DOOM) is around 5% or 10%", "source_url": "https://x.com/ManacasterBen/status/1886445455392776425", "source_type": "x", "date": "2025-02-03", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 278 }, { "name": "David Wood", "role": "Chair, London Futurists; Executive Director, LEV Foundation", "org": null, "stake": "independent", "low": 40.0, "high": 40.0, "label": "up to 40%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) may be as high as 40% these days", "source_url": "https://x.com/dw2/status/1889272788940124637", "source_type": "x", "date": "2025-02-11", "verified": true, "tags": [ "x-net-backfill" ], "note": "Phrased as 'may be as high as 40%'.", "id": 279 }, { "name": "Kitten", "role": "Pseudonymous account (@kitten_beloved)", "org": null, "stake": "independent", "low": 10.0, "high": 100.0, "label": "≥10%", "scope": "p(doom) mainly via AI-enabled misuse by humans", "horizon": null, "quote": "My probability of doom is probably 10% or higher, but not because we lose control of an AI, because it enables a hostile human actor to do very destructive things", "source_url": "https://x.com/kitten_beloved/status/1889472849754333240", "source_type": "x", "date": "2025-02-12", "verified": true, "tags": [ "x-net-backfill", "pseudonym" ], "note": null, "id": 280 }, { "name": "Max Spero", "role": "Co-founder, Pangram Labs ('slop janitor @pangram')", "org": null, "stake": "ai-industry", "low": 0.0, "high": 1.0, "label": "≪1%", "scope": "p(doom)", "horizon": null, "quote": "but also my current p doom is <<0.01", "source_url": "https://x.com/max_spero_/status/1889512443799347445", "source_type": "x", "date": "2025-02-12", "verified": true, "tags": [ "x-net-backfill" ], "note": "Given as a fraction (same post uses 0.8), i.e. well under 1%.", "id": 281 }, { "name": "Steve Phillips", "role": "Privacy hacker, DEF CON speaker", "org": null, "stake": "independent", "low": 80.0, "high": 80.0, "label": "80%", "scope": "p(doom), AI risk", "horizon": null, "quote": "my p(doom) is 80%", "source_url": "https://x.com/elimisteve/status/1898492298025222488", "source_type": "x", "date": "2025-03-08", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 282 }, { "name": "Lyman Stone", "role": "Demographer", "org": null, "stake": "academic", "low": 0.0, "high": 1.0, "label": "≈0%", "scope": "p(doom)", "horizon": null, "quote": "my p(doom) on ai is tbh basically zero", "source_url": "https://x.com/lymanstoneky/status/1912978073697526070", "source_type": "x", "date": "2025-04-17", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 283 }, { "name": "Greg Colbourn", "role": "Founder, CEEALAR; Pause AI advocate", "org": null, "stake": "safety-org", "low": 60.0, "high": 60.0, "label": "~60%", "scope": "p(doom)", "horizon": null, "quote": "That's why my overall p(doom) is only ~60% (I used to say 50%, before things went so badly in Paris).", "source_url": "https://x.com/gcolbourn/status/1923362212497813657", "source_type": "x", "date": "2025-05-16", "verified": true, "tags": [ "x-net-backfill" ], "note": "Up from 50%.", "id": 284 }, { "name": "David Manheim", "role": "Founder, ALTER; superforecaster emeritus", "org": null, "stake": "safety-org", "low": 50.0, "high": 50.0, "label": "~50%", "scope": "p(doom)", "horizon": null, "quote": "which is why my overall p(doom) is closer to 50%.", "source_url": "https://x.com/davidmanheim/status/1924022573332869293", "source_type": "x", "date": "2025-05-18", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 285 }, { "name": "Jey Grayson", "role": "ASI safety & alignment advocate", "org": null, "stake": "independent", "low": 70.0, "high": 70.0, "label": "70%", "scope": "p(doom)", "horizon": null, "quote": "My P(doom) or estimate on humanity extinction rate is at 70%!", "source_url": "https://x.com/JeyHiroGrayson/status/1938119919045919230", "source_type": "x", "date": "2025-06-26", "verified": true, "tags": [ "x-net-backfill" ], "note": "Existing row 90% (2026-09-20); earlier number.", "id": 286 }, { "name": "Greg Colbourn", "role": "Founder of Moratorium House and CEEALAR (EA Hotel); PauseAI advocate", "org": null, "stake": "safety-org", "low": 90.0, "high": 90.0, "label": "~90%", "scope": "p(doom)", "horizon": null, "quote": "I think p(doom) is only ~90%", "source_url": "https://x.com/gcolbourn/status/1941113482042331273", "source_type": "x", "date": "2025-07-04", "verified": true, "tags": [ "x-net-backfill" ], "note": "Said in a thread about Yampolskiy's p(doom|ASI); the remaining ~10% is simulation/aliens-type 'deus ex machina'.", "id": 287 }, { "name": "Phil Libin", "role": "Co-founder and CEO of All Turtles and mmhmm; co-founder and former CEO of Evernote", "org": null, "stake": "ai-industry", "low": 0.0, "high": 0.0, "label": "≈0%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) remains close to zero.", "source_url": "https://x.com/plibin/status/1945497479287329219", "source_type": "x", "date": "2025-07-16", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 288 }, { "name": "David Veksler", "role": "Bitcoin advocate, writer", "org": null, "stake": "independent", "low": 50.0, "high": 50.0, "label": "50%", "scope": "near-term human extinction from misaligned AI", "horizon": "near-term", "quote": "My p-doom (probability of near-term human extinction from misaligned AI) is 50%.", "source_url": "https://x.com/HeroicLife/status/1957946899828470005", "source_type": "x", "date": "2025-08-19", "verified": true, "tags": [ "x-net-backfill" ], "note": "Would be 90%+ if not a moral realist.", "id": 289 }, { "name": "Magnus Hambleton", "role": "VC at byFounders", "org": null, "stake": "investor", "low": 2.0, "high": 3.0, "label": "2-3%", "scope": "p(doom)", "horizon": null, "quote": "my p(doom) has gone from maybe 10% to 2-3%", "source_url": "https://x.com/magnushambleton/status/1971626911110779008", "source_type": "x", "date": "2025-09-26", "verified": true, "tags": [ "x-net-backfill" ], "note": "Down from ~10%.", "id": 290 }, { "name": "Zy", "role": "X commentator", "org": null, "stake": "independent", "low": 0.2, "high": 0.2, "label": "0.2%", "scope": "p(doom) from AI", "horizon": null, "quote": "my p(doom) from AI is .2%", "source_url": "https://x.com/ZyMazza/status/1978811408151130156", "source_type": "x", "date": "2025-10-16", "verified": true, "tags": [ "x-net-backfill", "pseudonym" ], "note": "Separately gives 60% for doom from all causes.", "id": 291 }, { "name": "Jey Grayson", "role": "ASI safety & alignment enthusiast", "org": null, "stake": "independent", "low": 99.1, "high": 99.1, "label": "99.1%", "scope": "p(doom) given ASI emergence", "horizon": null, "quote": "So my p(doom) for ASI emergence is 99.1%!", "source_url": "https://x.com/JeyHiroGrayson/status/1983589233550922105", "source_type": "x", "date": "2025-10-29", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 292 }, { "name": "AI Notkilleveryoneism Memes", "role": "AI-safety meme account", "org": null, "stake": "independent", "low": 25.0, "high": 25.0, "label": "~25%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) is only ~25%", "source_url": "https://x.com/AISafetyMemes/status/1991496216551780695", "source_type": "x", "date": "2025-11-20", "verified": true, "tags": [ "x-net-backfill", "pseudonym" ], "note": "Earlier: 26% (2023-11-26, 2024-07-28) and 27% (2023-12-07).", "id": 293 }, { "name": "vie (@viemccoy)", "role": "Researcher at OpenAI", "org": null, "stake": "lab", "low": 0.0, "high": 1.0, "label": "≈0%", "scope": "p(doom)", "horizon": null, "quote": "even if my p(doom) was on the extreme side (its functionally zero)", "source_url": "https://x.com/viemccoy/status/1991966813543108651", "source_type": "x", "date": "2025-11-21", "verified": true, "tags": [ "x-net-backfill", "pseudonym" ], "note": null, "id": 294 }, { "name": "Nikola Jurkovic", "role": "AI safety researcher, METR", "org": null, "stake": "safety-org", "low": 50.0, "high": 100.0, "label": ">50%", "scope": "p(doom) from AI", "horizon": null, "quote": "My P(doom) from AI is more than 50%", "source_url": "https://x.com/nikolaj2030/status/1992612415834603770", "source_type": "x", "date": "2025-11-23", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 295 }, { "name": "Bill Karr", "role": "Open Insights; PhD in Lorentzian geometry", "org": null, "stake": "ai-industry", "low": 0.0, "high": 0.0, "label": "≈0%", "scope": "p(doom)", "horizon": null, "quote": "my p(doom) is zero", "source_url": "https://x.com/billardkarr/status/2023867394226364829", "source_type": "x", "date": "2026-02-17", "verified": true, "tags": [ "x-net-backfill" ], "note": null, "id": 296 }, { "name": "Jack Feynman", "role": "Ex-energy operations and research (DoE, IHS Markit, DoD, Shell)", "org": null, "stake": "independent", "low": 30.0, "high": 30.0, "label": "30%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) is now 30%.", "source_url": "https://x.com/JackFeynman/status/2026008870296977618", "source_type": "x", "date": "2026-02-23", "verified": true, "tags": [ "x-net-backfill" ], "note": "Existing row is 0 (2026-09-13); this earlier post (2026-02-23) says 30%.", "id": 297 }, { "name": "Seán Ó hÉigeartaigh", "role": "Programme director at the University of Cambridge; researches AI and emerging-tech risks", "org": null, "stake": "academic", "low": 0.0, "high": 50.0, "label": "<50%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom)'s always been lower than 0.5", "source_url": "https://x.com/S_OhEigeartaigh/status/2031804764309144038", "source_type": "x", "date": "2026-03-11", "verified": true, "tags": [ "x-net-backfill" ], "note": "Given as a fraction (0.5 = 50%). His 'facetious' caveat refers to a separate joke, not the number.", "id": 298 }, { "name": "Rafael Ruiz de Lira", "role": "PhD candidate in philosophy, LSE", "org": null, "stake": "academic", "low": 10.0, "high": 10.0, "label": "~10%", "scope": "p(doom)", "horizon": null, "quote": "keeping my p(doom) around like 10%", "source_url": "https://x.com/RafaRuizdeLira/status/2037237868548984939", "source_type": "x", "date": "2026-03-26", "verified": true, "tags": [ "x-net-backfill" ], "note": "Existing row is 10–30% (2026-09-20); this earlier post (2026-03-26) says ~10%.", "id": 299 }, { "name": "somewheresy", "role": "Engineer (pseudonymous, 29k followers)", "org": null, "stake": "independent", "low": 0.0, "high": 0.0, "label": "0%", "scope": "p(doom)", "horizon": null, "quote": "My p(doom) is 0", "source_url": "https://x.com/somewheresy/status/2059763866830029134", "source_type": "x", "date": "2026-05-27", "verified": true, "tags": [ "x-net-backfill", "pseudonym" ], "note": "Established pseudonym; bare statement, no context.", "id": 300 } ], "declined": [ { "name": "Donald Trump", "role": "US President", "quote": "I am not worried at all. [...] What I worry about is that, if we do not win the AI race, we will be placed in a very disadvantageous position.", "date": "2026-09-10", "source_url": "https://www.khan.co.kr/en/article/202609111612007/", "note": "Asked by reporters whether AI could bring about the extinction of humanity." }, { "name": "Ted Cruz", "role": "US Senator; Commerce Committee chair", "quote": "if they're going to be killer robots, I'd rather they be American killer robots and not Chinese killer robots.", "date": "2026-09-09", "source_url": "https://www.huffpost.com/entry/ted-cruz-ai-american-killer-robots_n_6aa337f3e4b0b2af35a5a08d", "note": "Asked by Politico about AI extinction warnings; gave no number of his own." }, { "name": "Stuart Russell", "role": "Professor of computer science, UC Berkeley", "quote": "what's your probability of doom, your so-called P of doom number? Because that makes sense if you're an alien... if you're a human, then you're not just betting, you're actually acting.", "date": "2025-12-04", "source_url": "https://singjupost.com/stuart-russell-on-the-diary-of-a-ceo-podcast-transcript/", "note": "On The Diary of a CEO; relays other people's numbers but gives none of his own." }, { "name": "Neel Nanda", "role": "Mechanistic interpretability lead, Google DeepMind", "quote": "I have a policy that I just refuse to answer in public what my p(doom) or timelines are, because I think people just anchor too much on what someone prominent says", "date": "2025-09-15", "source_url": "https://80000hours.org/podcast/episodes/neel-nanda-career-advice-frontier-ai-companies/", "note": "Standing policy stated on the 80,000 Hours Podcast." }, { "name": "Demis Hassabis", "role": "CEO, Google DeepMind", "quote": "I don't have a P-Doom number. The reason I don't is because I think it would imply a level of precision that is not there... it's definitely non-zero and it's probably non-negligible.", "date": "2025-07-23", "source_url": "https://lexfridman.com/demis-hassabis-2-transcript/", "note": "Asked by Lex Fridman." }, { "name": "Sundar Pichai", "role": "CEO, Google and Alphabet", "quote": "I think I'm optimistic on the p(doom) scenarios, but that doesn't mean I think the underlying risk is actually pretty high.", "date": "2025-06-05", "source_url": "https://lexfridman.com/sundar-pichai-transcript/", "note": "Asked by Lex Fridman, who gave his own 10%; Pichai gave no number. Other lists attribute Lex's 10% to him." }, { "name": "Max Harms", "role": "Researcher, MIRI", "quote": "When I try to tell a concrete story of the development of superintelligence going well ... I simply fail. In that sense, my P(doom) is 100%.", "date": "2025", "source_url": "https://www.lesswrong.com/posts/6h7aAd4aw8YgCAbF6/p-doom-is-a-dumb-meme", "note": "Gives an inside-view 100% but refuses an all-things-considered number; calls p(doom) 'a dumb meme'." }, { "name": "Guillaume Verdon (Beff Jezos)", "role": "Founder, Extropic; e/acc co-founder", "quote": "I'm not a fan of that calculation, I think people just throw numbers out there and it's a very sloppy calculation, right?", "date": "2023-12-29", "source_url": "https://lexfridman.com/guillaume-verdon-transcript", "note": "Asked by Lex Fridman." }, { "name": "Robert Miles", "role": "AI safety YouTuber", "quote": "I don't know. I think it doesn't look good. It's fair to say it doesn't look good. My mainline prediction is doom.", "date": "2022-08-17", "source_url": "https://theinsideview.ai/rob", "note": "Asked on The Inside View; gave no number, but puts 'maybe 10% or 20%' on being wrong in the lucky direction." }, { "name": "Eliezer Yudkowsky", "role": "Co-founder, MIRI", "quote": null, "date": null, "source_url": "https://www.lesswrong.com/posts/j9Q8bRmwCgXRYAgcJ/miri-announces-new-death-with-dignity-strategy", "note": "Long-standing practice of not giving a doom number; the '>95%' on other lists has no primary source. Writes of dying 'with dignity' rather than a percentage." }, { "name": "Anatoly Yakovenko (toly)", "role": "Solana co-founder", "quote": "My p(doom) is 100% if humans don’t develop asi", "date": "2026-09-26", "source_url": "https://x.com/toly/status/2103988609737199943", "note": "Asked on X \"so are u team p(doom) toly\"; answered with the risk of not building ASI, not of AI itself. No number for AI-caused doom, so not a row." }, { "name": "Sam Altman", "role": "CEO, OpenAI", "quote": "I have never known how to put an exact number on p(doom), to say nothing of how to think about the differential. I will say this: if I thought going to work every day made it less likely that we all continue to thrive into the future, I would retire and just hang with my kids.", "date": "2026-03-01", "source_url": "https://x.com/sama/status/2027958834741055691", "note": "Reply to Geoffrey Miller (@gmiller) asking 'what's the maximum p(doom) you're willing to impose on all of our kids'. Verified via syndication API. Wikipedia's '>0%' for Altman is words, not a number." }, { "name": "Kenneth Stanley", "role": "Former OpenAI research team lead (open-endedness)", "quote": "And I think if I give you a p-doom, I will immediately polarize your audience. ... I just don't really want to be pigeonholed into this number.", "date": "2025-01-06", "source_url": "https://lironshapira.substack.com/p/debate-with-a-former-openai-research", "note": "Asked on Doom Debates (~58:30-59:05, auto-transcript). Stake ex-lab. He also says a 50% p(doom) is 'within reason' given his wide error bars, which is not his own number." }, { "name": "Alex Bores", "role": "New York State Assemblymember; Democratic candidate for Congress (NY-12); RAISE Act author", "quote": "Ben Smith: Like 5%, 10%? Alex Bores: I wouldn't put a number on it, but I think a lot of the debates on it, is it 5%, is it 10%, is it 50%, is it 90%? If your chance of human extinction is measurable in percentage points, the federal government should be thinking about it and acting on it. Ben Smith: Do you think it is in integers? Alex Bores: Yes. The median AI researcher, when polled, puts the risk of extinction at 10%. Ben Smith: Do you think that's reasonable? Alex Bores: I think so.", "date": "2026-05-13", "source_url": "https://nyeditorialboard.substack.com/p/alex-bores-on-housing-ai-regulation", "note": "Said he \"wouldn't put a number on it\"; agreed it is measurable in whole percentage points and that 10% is reasonable. Not recorded as a number." }, { "name": "Alexandria Ocasio-Cortez", "role": "US Representative (D-NY)", "quote": "I just don't think it's constructive to engage in it", "date": "2026-09-18", "source_url": "https://www.politico.com/newsletters/playbook/2026/09/18/lina-khans-cure-for-dem-ai-doomers-01083630", "note": "Asked by POLITICO Playbook for her P(doom). Playbook: she 'knew the score. But she wasn't playing the numbers game.'" }, { "name": "Ruben Gallego", "role": "US Senator (D-AZ)", "quote": "This doesn't end in us like having a nuclear apocalypse. But what happens when an agent goes rogue and actually wipes out everybody's banking history?", "date": "2026-09-18", "source_url": "https://www.politico.com/newsletters/playbook/2026/09/18/lina-khans-cure-for-dem-ai-doomers-01083630", "note": "Answered POLITICO Playbook's P(doom) question with no number; implies low extinction risk in words only." }, { "name": "JB Pritzker", "role": "Governor of Illinois", "quote": "No one should be asking politicians to calculate mathematical probabilities related to AI. Instead, listen to what the experts are telling the world: AI is a threat to humanity's existence.", "date": "2026-09-18", "source_url": "https://www.politico.com/newsletters/playbook/2026/09/18/lina-khans-cure-for-dem-ai-doomers-01083630", "note": "Statement from a Pritzker spokesperson, not Pritzker in person." }, { "name": "Hodan Omaar", "role": "Senior Policy Analyst, ITIF", "quote": "The way I think about it is that my range for what p(doom) for AGI is is somewhere around what my p(doom) for other hypothetical, but yet unproven, risks are, rather than necessarily putting a number on it", "date": "2023-12-06", "source_url": "https://www.techpolicy.press/us-senate-ai-insight-forum-tracker/", "note": "Participant in Senate AI Insight Forum #8, where Schumer asked each attendee for their p(doom). Quote from her interview with Tech Policy Press. Think-tank (independent), not government." }, { "name": "Audrey Tang", "role": "Taiwan's Cyber Ambassador-at-Large; former Digital Minister", "quote": "My P(Doom) is NaN, or Not A Number, and here's why. [...] It is from zero percent to one hundred percent. We're flying blind. We don't know. So the mean, I guess, is fifty percent, but it's cosmetic. The interval is the message. [...] It's zero to one hundred. [...] we need a glass horse system that you can actually see into as a condition of deployment, and then I'll give you a real number.", "date": "2026-01-20", "source_url": "https://lironshapira.substack.com/p/can-we-govern-superintelligence-taiwans", "note": "Doom Debates transcript, 0:17:34 to 0:20:43. Refuses a number ('NaN', '0 to 100'). The '50% mean' is explicitly 'cosmetic', so not a row." }, { "name": "Shaw Walters", "role": "Founder, Eliza Labs / elizaOS (formerly ai16z)", "quote": "So what's your number for P(doom)? I think this is a certain flaw in relatively smart people that they feel this need to take a hyper complex vector and turn it into a scalar number. ... I just don't believe in doom. ... My P(doom) is a multipolar. It's gotta be at least 10 dimensions at this point. It's pretty low on some of those dimensions, but mostly it's pretty high.", "date": "2026-03-06", "source_url": "https://www.blockchaingamer.biz/features/interviews/41648/shaw-walters-elizaos-ai-agents-blockchain-gaming/", "note": "BlockchainGamer.biz interview by Jon Jordan. Asked directly for a number, gave none ('mostly it's pretty high' is words, not a number). Crypto-AI agent founder." }, { "name": "Casey Muratori", "role": "Game developer, programming educator (Molly Rocket), co-host of Wading Through AI", "quote": "I hate to disappoint you. As with the things that I said before, I don't think I can actually put a probability on it because of the inability to know all of the specifics of what people are actually working on.", "date": "2026-06-25", "source_url": "https://lironshapira.substack.com/p/even-01-pdoom-is-unacceptable-world", "note": "Doom Debates ~55:10, asked 'what is your P-Doom'. Episode title 'Even 0.1% P(Doom) Is Unacceptable' is his threshold argument, not his estimate. Quote from Substack auto-transcript. Tech-twitter/YouTube figure, not crypto." }, { "name": "Rohan Anil", "role": "Co-founder, Core Automation; formerly Anthropic and Google DeepMind", "quote": "No, my p(doom) is not 10%", "date": "2026-09-10", "source_url": "https://x.com/_arohan_/status/2097874723195805748", "note": "Opening an AMA after leaving Anthropic; says what his number isn't, not what it is." } ], "stake_legend": { "lab": "employed by or founded a frontier lab (OpenAI, Anthropic, Google DeepMind, Meta AI, xAI) at the time of the statement", "ex-lab": "formerly at a frontier lab", "ai-industry": "builds or sells AI outside the frontier labs (chips, models, startups)", "investor": "invests in AI companies", "safety-org": "works at an AI safety or x-risk organisation", "academic": "university researcher", "government": "government or policy role", "media": "journalist, podcaster, writer or streamer", "independent": "none of the above publicly known" } }