Scores by kind of claim
Each kind of claim is scored on its own, out of 100: the share that held up, with the range it likely falls in. Fewer resolved claims give a wider range.
-
Factual claims
Statements about the present or the past, checked against the record: did they hold up?
76 70 82 -
Predictions
Statements about events outside the person's control, checked once they fell due: did they come true?
60 42 78
All types combined: 74 (likely 68 to 80, high confidence), from 79 claims: 71 factual claims, 8 predictions. It mixes kinds of claim in different proportions for different people, so compare people kind by kind.
Eliezer Yudkowsky's claims
21 claims
“Alignment will not be solved in a few years.”
Eliezer Yudkowsky, interviewee 6 April 2023 Eliezer Yudkowsky - Why AI Will Kill Us, Aligning LLMs, Nature of Intelligence, SciFi, & Rationality (Dwarkesh Podcast)
AI alignment will not be solved within a few years.
Resolved true The record shows it happened or was so.
The second judge would not resolve this claim; the verdict shown is the first judge's.
What would count as true or false
'A few years' is taken as about three years from April 2023. True: by then there is no broad expert consensus, among AI safety researchers and leading labs, that the alignment problem (reliably aligning advanced or superhuman AI) has been solved. Mostly true: no consensus exists, but some lab or researcher has claimed a solution that is widely disputed. Mixed: a significant share of experts consider alignment largely solved. Mostly false: most experts accept a claimed solution, with notable dissent. False: there is broad expert and lab consensus that alignment of advanced AI has been solved.
Evidence
- AI alignment is solvable / YouCongress YouCongress · 22 October 2026
“Fundamentally, we just do not know how to do alignment. If somebody came with a thousand-page document that explained exactly how to do everything, we couldn’t get a model to learn exactly that.”
- AI alignment is solvable / YouCongress YouCongress · 3 November 2026
“Many AI researchers grant these concerns and recognize that the hard version of the alignment problem is unsolved; however, they generally believe that the better models of the future will help solve it.”
- Automated researchers can reliably mitigate alignment failures Anthropic · 28 August 2026
“Overall, we view these results as early positive signals that automated alignment post-training could become practical in the near term, and we will share updates as this work progresses.”
- Survey of AI safety leaders on x-risk, AGI timelines, and resource allocation (Feb 2026) EA Forum · 25 March 2026
“Current safety frameworks and evaluation practices are necessary but insufficient.”
Reasoning
A survey of AI safety leaders conducted in February 2026 reports that current safety frameworks and evaluation practices are "necessary but insufficient" (E4). Later statements from AI safety researchers say alignment is still not known how to do (E1) and that the hard version of the problem is unsolved (E2). In August 2026 Anthropic described its automated alignment work only as "early positive signals" (E3). None of this points to a broad expert or lab consensus that alignment of advanced AI was solved by April 2026, so the prediction resolves true.
Cite this claim
Eliezer Yudkowsky, “AI alignment will not be solved within a few years.” (said 6 April 2023). Verdict: resolved true. Reliability API, claim c_879fc584608e, method 0.1.0+c876030c. https://epistacklabs.com/claims/c_879fc584608e
Claims without a verdict
Is a verdict here wrong, or do you speak for Eliezer Yudkowsky? Reply on the claim's own page, with the form at its foot (how replies work). Verdicts are made automatically; how verdicts and scores are made.