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
42 claims
“GPT-5 by itself won’t be a general purpose artificial general intelligence capable of taking on arbitrary tasks.”
Gary Marcus, author 27 July 2025 What to Expect When You're Expecting … GPT-5 (Marcus on AI)
GPT-5 by itself will not be a general-purpose AGI capable of arbitrary tasks.
Resolved true The record shows it happened or was so.
What would count as true or false
The claim is that GPT-5 on its own would not be a general-purpose AGI able to take on arbitrary tasks. True: after GPT-5's release, OpenAI does not credibly claim it is AGI, and the expert consensus and benchmarks or evaluations show GPT-5 alone cannot reliably do arbitrary tasks. Evidence would include clear failures on broad agentic or novel tasks, ARC-AGI style tests, or long-horizon real-world tasks. Mostly true: the same, but a noticeable minority of credible experts argue it meets some definition of AGI. Mixed: credible expert opinion is genuinely split. Mostly false: OpenAI or a broad expert consensus treats GPT-5 alone as AGI, with only narrow exceptions. False: GPT-5 by itself is widely shown and accepted to perform arbitrary tasks at a general human level.
Evidence
- Details about METR's evaluation of OpenAI GPT-5 METR · 7 August 2025
“To determine if GPT-5 is capable of executing any of our three main threat models, we used our time-horizon evaluation and determined it has a 50% time-horizon of around 2 hours 17 minutes on agentic software engineering tasks.”
- Details about METR's evaluation of OpenAI GPT-5 METR · 7 August 2025
“Like other models, GPT-5 performed poorly on some of the longest HCAST tasks, and some shorter tasks which load particularly heavily on forward planning.”
- OpenAI's GPT-5 is here | TechCrunch TechCrunch · 7 August 2025
“During a briefing with reporters, OpenAI CEO Sam Altman claimed GPT-5 is “the best model in the world,” and said it represented a “significant step” along the company’s path to developing AI that can outperform humans at most economically valuable work — that is, artificial general intelligence (AGI).”
- OpenAI Researcher Mocks Elon Musk's AGI Claim to His Face Futurism · 21 October 2025
“In the build up to the launch of OpenAI’s newest GPT-5 model this summer — considered a massive disappointment by many — Altman declared it a “significant step along the path to AGI.””
Reasoning
At launch, OpenAI's CEO called GPT-5 a "significant step" toward AGI rather than AGI itself (E3, E4). Neither source reports OpenAI claiming that GPT-5 is AGI. METR's independent evaluation measured a 50% time horizon of about 2 hours 17 minutes on agentic software tasks, and found that GPT-5 did poorly on some of the longest and most planning-heavy tasks (E1, E2). The evidence shows limits on long-horizon and planning-heavy tasks [E1][E2], and none of it reports OpenAI or credible experts treating GPT-5 alone as AGI.
Cite this claim
Gary Marcus, “GPT-5 by itself will not be a general-purpose AGI capable of arbitrary tasks.” (said 27 July 2025). Verdict: resolved true. Reliability API, claim c_a05902e075bc, method 0.1.0+c876030c. https://epistacklabs.com/claims/c_a05902e075bc
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.