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?
79 72 85 -
Predictions
Statements about events outside the person's control, checked once they fell due: did they come true?
80 72 88 -
Promises
Commitments to the person's or organisation's own future action, with something to deliver: was it delivered?
Insufficient record: 1 promise resolved, 5 needed for a score.
All types combined: 79 (likely 74 to 84, high confidence), from 102 claims: 64 factual claims, 38 predictions. It mixes kinds of claim in different proportions for different people, so compare people kind by kind.
Gary Marcus's claims
52 claims
“certain people are hard at work denying that there is any AI risk at all, either long term or short, such as Marc Andreessen, trying to turn the question of whether to pursue AI safety at all into a debate.”
Gary Marcus, author of Marcus on AI 26 October 2023 A CERN for AI and the Global Governance of AI (Marcus on AI)
Some people, such as Marc Andreessen, are denying that there is any AI risk, long term or short term, and trying to turn whether to pursue AI safety into a debate.
Mostly true True in substance, wrong in a detail.
What would count as true or false
Judge as of 2023-10-26. True: before that date, Marc Andreessen (the named example) publicly denied that AI poses any meaningful risk, both long-term (existential) and short-term (for example misinformation or bias), and framed the question of whether to pursue AI safety at all as a matter of debate. Evidence could include his June 2023 essay 'Why AI Will Save the World' or his October 2023 'Techno-Optimist Manifesto'. Mostly true: Andreessen dismissed most AI risks, both long- and short-term, and opposed safety efforts, but acknowledged some limited risk (for example misuse by bad actors), so 'any AI risk at all' is a modest overstatement. Mixed: he dismissed long-term (existential) risk but clearly acknowledged a range of short-term risks, or the reverse. Mostly false: his statements mainly argue the benefits of AI or oppose regulation while acknowledging significant risks. False: he had publicly acknowledged substantial AI risks and supported safety work.
Evidence
- Marc Andreessen on AI safety and AI for the world - Marginal REVOLUTION Marginal Revolution (Tyler Cowen) · 6 June 2023
“So far I have explained why four of the five most often proposed risks of AI are not actually real – AI will not come to life and kill us, AI will not ruin our society, AI will not cause mass unemployment, and AI will not cause an ruinous increase in inequality.”
- Why AI Will Save The World Marc Andreessen (pmarca Substack) · 6 June 2023 Own record
“But now let’s address the fifth, the one I actually agree with: AI will make it easier for bad people to do bad things.”
- The Techno-Optimist Manifesto | Andreessen Horowitz Andreessen Horowitz · 16 October 2023 Own record
“Our present society has been subjected to a mass demoralization campaign for six decades – against technology and against life – under varying names like “existential risk”, “sustainability”, “ESG”, “Sustainable Development Goals”, “social responsibility”, “stakeholder capitalism”, “Precautionary Principle”, “trust and safety”, “tech ethics”, “risk management”, “de-growth”, “the limits of growth”.”
- Silicon Valley rushes to save the case for techno-optimism after Marc Andreessen’s hallucination of a manifesto | Fortune Fortune · 20 October 2023
“Andreessen says trust and safety measures are the “enemy,” whereas Altman says safety can’t be determined in a lab but must be influenced by public use.”
Reasoning
Tyler Cowen's excerpt of Andreessen's June 2023 essay (E1) has Andreessen arguing that four of the five most often proposed AI risks are "not actually real". These include existential risk ("kill us") and societal harms ("ruin our society"). In the same essay (E2), though, he says he agrees with the fifth risk, that AI will make it easier for bad people to do bad things. So the claim that he denies "any AI risk at all" overstates his position somewhat. His October 2023 manifesto (E3) and Fortune's coverage of it (E4) show him casting "existential risk" and "trust and safety" as targets or enemies, which fits the rest of the claim. Together this matches the "mostly true" criterion.
Cite this claim
Gary Marcus, “Some people, such as Marc Andreessen, are denying that there is any AI risk, long term or short term, and trying to turn whether to pursue AI safety into a debate.” (said 26 October 2023). Verdict: mostly true. Reliability API, claim c_1e401fe7fda6, method 0.1.0+c876030c. https://epistacklabs.com/claims/c_1e401fe7fda6
Claims without a verdict
Is a verdict here wrong, or do you speak for Gary Marcus? 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.