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?
72 67 78 -
Predictions
Statements about events outside the person's control, checked once they fell due: did they come true?
Insufficient record: 1 prediction resolved, 5 needed for a score.
-
Promises
Commitments to the person's or organisation's own future action, with something to deliver: was it delivered?
Insufficient record: 4 promises resolved, 5 needed for a score.
-
Guidance
A company's own figures for its results over a period, checked against what it then reported.
Insufficient record: 1 guidance claim resolved, 5 needed for a score.
All types combined: 72 (likely 67 to 78, high confidence), from 101 claims: 101 factual claims. It mixes kinds of claim in different proportions for different people, so compare people kind by kind.
Jensen Huang's claims
49 claims
“The longer you think, the better the quality answer you get.”
Jensen Huang, Nvidia CEO 26 September 2025 Transcript: Jensen Huang on OpenAI, Future of Compute (BG2 Pod) (Singju Post)
AI models that think longer (inference-time reasoning) produce better quality answers.
Mostly true True in substance, wrong in a detail.
What would count as true or false
Judged as of 2025-09-26, as a general claim about inference-time reasoning. TRUE: published research and model reports (for example OpenAI o1/o3 reports, DeepSeek-R1 and test-time compute scaling papers) consistently show that more inference-time reasoning or compute improves answer quality or accuracy, broadly and not just on narrow tasks. MOSTLY TRUE: research shows a general positive relation but documents important limits, such as diminishing returns, 'overthinking' cases where longer reasoning hurts, or gains mainly on reasoning-heavy tasks, so the unqualified 'the longer, the better' overstates a real trend. MIXED: the evidence is evenly split. MOSTLY FALSE: improvements appear only in narrow settings and the evidence is otherwise inconclusive. FALSE: research shows that longer inference-time reasoning generally does not improve answer quality.
Evidence
- DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning - Nature Nature (DeepSeek-AI authors) · 17 September 2025
“Nevertheless, there remains room for further optimization in terms of token efficiency, as instances of excessive reasoning—manifested as overthinking—are still observed in response to simpler questions.”
- [PDF] Inference-Time Scaling for Complex Tasks - Microsoft Microsoft Research · 3 February 2026
“Longer generation is not always better. For example, DeepSeek R1 consumes tokens 10 times more than Claude 3.7 Sonnet (Figure 4), but its accuracy is even slightly lower.”
- Revisiting the Test-Time Scaling of o1-like Models: Do They Truly Possess Test-Time Scaling Capabilities? Association for Computational Linguistics (ACL 2025, Long Papers) · 31 July 2026
“ThisstudyfoundthatlongerCoTsoftheseo1-likemodelsdonotconsistentlyenhanceaccuracy;infact,correctsolutionsareoftenshorterthanincorrectonesforthesamequestions.”
- Scaling LLM Test-Time Compute Optimally Can be More Effective than Scaling Parameters for Reasoning ICLR 2025 (oral)
“We find that in both cases, the effectiveness of different approaches to scaling test-time compute critically varies depending on the difficulty of the prompt.”
Reasoning
E4 (ICLR 2025) reports that scaling test-time compute can work better than scaling parameters on reasoning tasks, but that how well it works depends on how hard the prompt is. E1 (DeepSeek-R1 in Nature) supports reasoning-based gains but notes that the model still overthinks simpler questions. E2 (Microsoft Research) finds that longer generation is not always better, and E3 (ACL 2025) finds that longer chains of thought in o1-like models do not consistently raise accuracy. Taken together, the evidence shows a real positive relation with important limits, so the unqualified "the longer, the better" overstates it.
Cite this claim
Jensen Huang, “AI models that think longer (inference-time reasoning) produce better quality answers.” (said 26 September 2025). Verdict: mostly true. Reliability API, claim c_6658d8e5a5c5, method 0.1.0+03d23c0e. https://epistacklabs.com/claims/c_6658d8e5a5c5
Claims without a verdict
Is a verdict here wrong, or do you speak for Jensen Huang? 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.