p(doom) board
Every p(doom) on the record.
p(doom) is shorthand for the probability a person assigns to civilisational catastrophe from AI. Definitions vary: extinction, disempowerment, loss of control, or just bad outcomes. The claim only means what the person said it means, on the date they said it. Every entry below links to its source.
distribution · latest claim per person
- ≥ 90%2
- 50–89%8
- 20–49%8
- 10–19%7
- 1–9%2
- < 1%1
A person who has stated multiple p(doom) values shows up once here, using their most recent claim. Below, every claim is listed, including past ones, so a single person can appear multiple times.
mean p(doom) by vintage · era of formation
latest claim per person- Symbolic era38% · n=3
- Pre-deep-learning48% · n=4
- Deep-learning rise25% · n=3
- Scaling era41% · n=10
- Post-ChatGPT27% · n=5
The honest test of whether era predicts estimate. n is small per tier; read this as a signal, not a verdict. Tiers with fewer than 3 datapoints are hidden.
≥ 90%
2
Roman Yampolskiy
100%University of Louisville professor; argues AI safety is impossible
Explicit Twitter statement.
2024-03-13Tweet from Roman Yampolskiy · X/Twitter
Founder of MIRI; the original AI-extinction pessimist
Probability that AI wipes out humanity; Yudkowsky has repeatedly said >95%, sometimes framed as 99%.
2023PauseAI aggregated p(doom) list · PauseAI
50–89%
8Director of the Center for AI Safety; drafter of the Statement on AI Risk
Hendrycks has publicly indicated a p(doom) above 80%.
2023-04-02Tweet from Dan Hendrycks · X/Twitter
Former OpenAI governance team member; author of AI 2027 scenario
Self-reported ~70% chance of existential catastrophe.
2023LessWrong comment by Daniel Kokotajlo · LessWrong
Don't Worry About The Vase; weekly AI newsletter
2023-11-28Tweet reporting Zvi's p(doom) · X/Twitter
Holden Karnofsky
10–90%Co-founder of Open Philanthropy; AI safety funder and strategist
Explicitly unresolved wide range, framed around uncertainty about alignment difficulty.
2022-08How might we align transformative AI if it's developed very soon? · AI Alignment Forum
Jan Leike
10–90%Former head of OpenAI Superalignment; now at Anthropic
Large uncertainty range cited in interview.
2023-08Jan Leike interview (PauseAI citation) · YouTube
Former CEO of Stability AI; open-source frontier advocate
2024-12-04Tweet from Emad Mostaque · X/Twitter
Founder; Doom Debates podcast host
Existential catastrophe from AI in the next several decades.
2024Liron Shapira on Doom Debates · YouTube
UNM evolutionary psychologist; AGI pause advocate
~50% p(doom) with wide error bars (5–80%).
2024Modern Wisdom, Geoffrey Miller episode · Modern Wisdom
20–49%
9Founder of the US AI Safety Institute safety team; ex-OpenAI alignment lead
Approximately 46% chance of an extremely bad outcome, in his LessWrong post decomposing takeover and non-takeover catastrophes.
2023-04-27My views on doom · LessWrong
Eli Lifland
35%Forecaster; co-author of AI 2027
2023Eli Lifland on navigating the AI alignment landscape · EA Forum
Astral Codex Ten / Slate Star Codex blogger
2023-03-14Why I Am Not As Much Of A Doomer As Some People · Astral Codex Ten
Geoffrey Hinton
10–50%Godfather of deep learning; left Google in 2023 to speak about AI risk
Probability AI leads to human extinction in the next 30 years
2024-06PauseAI aggregated p(doom) list · PauseAI
Joseph Carlsmith
10–50%Open Philanthropy researcher; 'Is Power-Seeking AI an Existential Risk?'
Existential catastrophe from misaligned power-seeking AI by 2070; revised range.
2022Is Power-Seeking AI an Existential Risk? · Open Philanthropy
Emmett Shear
5–50%Former interim CEO of OpenAI; Twitch co-founder
2023-09Emmett Shear on AI risk · YouTube
Turing Award laureate; scientific chair of the International AI Safety Report
Probability of AI catastrophe (reported in ABC News piece).
2023-07-15What's your p(doom)? AI researchers worry catastrophe · ABC News
Reid Hoffman
20%LinkedIn co-founder; AI optimist investor
2024-09Future of AI (PBS) · PBS
Carl Shulman
20%Open Phil senior research analyst; AGI takeoff economics
Existential catastrophe from AI; rough rather than precise.
2023Carl Shulman on the moral status of AI · Dwarkesh Podcast
10–19%
7
Dario Amodei
10–25%CEO of Anthropic; 'Machines of Loving Grace' author
Publicly cited 10–25% chance of catastrophic outcomes.
2023-10PauseAI aggregated p(doom) list · PauseAI
Elon Musk
10–20%CEO of Tesla and xAI; co-founded OpenAI
Has stated 10–20% chance AI destroys humanity; in 2024 said 20%.
2024-04Elon Musk on AI risk (video) · YouTube
Lina Khan
15%Former chair of the FTC
2023-11Tweet citing Lina Khan's p(doom) · X/Twitter
Ethereum co-founder; author of 'My techno-optimism' manifesto
2023-11-28Tweet from Vitalik Buterin · X/Twitter
Toby Ord
10%Philosopher; author of The Precipice
Probability of existential catastrophe from unaligned AI in the next 100 years, as estimated in The Precipice.
2020-03-05The Precipice · Bloomsbury
Lex Fridman
10%MIT researcher; long-form podcast host
2024Lex Fridman Podcast · Lex Fridman Podcast
Open Philanthropy researcher; 'Is Power-Seeking AI an Existential Risk?'
Probability of existential catastrophe from misaligned AI by 2070.
2022-06-23Is Power-Seeking AI an Existential Risk? · arXiv
1–9%
2
Nate Silver
5–10%Statistician; Silver Bulletin / FiveThirtyEight founder
2024-08It's time to come to grips with AI · Silver Bulletin
Keen Technologies founder; ex-Meta CTO
Existential catastrophe from AI within his lifetime; rough.
2023John Carmack on Lex Fridman Podcast · Lex Fridman
< 1%
1Chief AI Scientist at Meta; outspoken AI-doom skeptic
LeCun repeatedly places his p(doom) effectively at zero; <0.01% reported by aggregator.
2023-12PauseAI aggregated p(doom) list · PauseAI
