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?
83 77 88 -
Predictions
Statements about events outside the person's control, checked once they fell due: did they come true?
69 49 87 -
Promises
Commitments to the person's or organisation's own future action, with something to deliver: was it delivered?
69 49 87
All types combined: 81 (likely 76 to 86, high confidence), from 87 claims: 77 factual claims, 5 predictions, 5 promises. It mixes kinds of claim in different proportions for different people, so compare people kind by kind.
Sam Altman's claims
4 claims
| Verdict | Kind | Said | Claim | Topic |
|---|---|---|---|---|
| Mixed | Factual | 11 Sep 2025 | Current AI models now hallucinate only a little bit. Factual · 11 Sep 2025 | AI capabilities |
| Mixed | Factual | 11 Sep 2025 | AI models such as ChatGPT do not do anything unless a user asks them to. Factual · 11 Sep 2025 | AI capabilities |
| Mixed | Factual | 9 Aug 2025 | GPT-5 is significantly better than earlier models at healthcare-related queries. Factual · 9 Aug 2025 | AI capabilities |
| Mixed | Factual | 16 Feb 2019 | There has been a massive surge in the availability of cheap seed funding for startups. Factual · 16 Feb 2019 | Economics and markets |
“GPT-5 is significantly better at healthcare related queries.”
Sam Altman, OpenAI CEO 9 August 2025 Transcript: Sam Altman Interview on Huge Conversations Podcast (Singju Post)
GPT-5 is significantly better than earlier models at healthcare-related queries.
Mixed Partly true and partly false in roughly equal measure.
The second judge said resolved true; the verdict shown is the first judge's.
Said for an organisation they lead or work for; it counts on both records.
What would count as true or false
True if benchmark results available around 2025-08-09 (e.g., HealthBench or HealthBench Hard, medical QA benchmarks, hallucination rates on health queries) from OpenAI's published evaluations or independent tests show that GPT-5 performed significantly better on healthcare-related queries than earlier OpenAI models such as GPT-4o and o3: a clear, material improvement, not a marginal one. Mostly true if GPT-5 improved clearly on most health measures but only modestly against the strongest earlier model (e.g., o3) on some. Mixed if the improvements were significant on some health benchmarks but negligible or reversed on others, or if OpenAI's figures show gains that independent evaluations do not confirm. Mostly false if the gains were only marginal or limited to a narrow sub-measure presented as general. False if the evidence shows GPT-5 performed no better or worse than earlier models on healthcare queries.
Evidence
- Introducing GPT-5 OpenAI · 7 August 2025 Own record
“The model scores significantly higher than any previous model on HealthBench”
- GPT-5 System Card - OpenAI Deployment Safety Hub OpenAI · 7 August 2025 Own record
“Errors in potentially urgent situations are reduced over 50x from GPT-4o and over 8x from OpenAI o3.”
- Clinical Accuracy and Safety Concerns Following GPT-5 Public Demonstration in Cancer Care PubMed Central (authors at University Hospital Tübingen, per data statement) · 29 November 2025
“Overall concordance with tumour board recommendations was 79/100 for GPT-4o-mini (95% CI 70.02–85.83%), and 85/100 for both GPT-4o and GPT-5”
- Clinical Accuracy and Safety Concerns Following GPT-5 Public Demonstration in Cancer Care PubMed Central (authors at University Hospital Tübingen, per data statement) · 29 November 2025
“found identical 85% concordance to GPT-4o, contradicting superiority claims.”
Reasoning
OpenAI's launch post says GPT-5 scores "significantly higher than any previous model on HealthBench" (E1). Its system card reports that errors in potentially urgent health situations fell more than 50x from GPT-4o and more than 8x from o3 (E2). Both are OpenAI's own evaluations. An independent study of 100 cancer-care cases found GPT-5 and GPT-4o equally in line with tumour-board recommendations, at 85/100 each, which runs against the claimed superiority (E3, E4). Under the criteria, gains in OpenAI's figures that an independent evaluation does not confirm count as mixed, though the independent study covers only one narrow clinical task.
Cite this claim
Sam Altman, “GPT-5 is significantly better than earlier models at healthcare-related queries.” (said 9 August 2025). Verdict: mixed. Reliability API, claim c_a0968b3dfb9d, method 0.1.0+c876030c. https://epistacklabs.com/claims/c_a0968b3dfb9d
Claims without a verdict
Is a verdict here wrong, or do you speak for Sam Altman? 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.