A lot of current AI models are very good at STEM.
Resolved true The record shows it happened or was so.
Factual: Statements about the present or the past, checked against the record: did they hold up?
The second judge would not resolve this claim; the verdict shown is the first judge's.
What would count as true or false
True if benchmark results and evaluations as of 2026-02-07 show that many leading AI models perform at a high level on STEM tasks, such as maths competitions, coding, science question sets like GPQA, or similar. Mostly true if the models are strong in most STEM areas but clearly weak in one notable area. Mixed if their performance is high in some STEM areas and poor in others in roughly equal measure. Mostly false if only a few top models perform well, or only on narrow benchmarks. False if current models generally perform poorly on STEM.
Set before the research began.
Evidence (4)
-
E1 Advanced version of Gemini with Deep Think officially achieves gold-medal standard at the International Mathematical Olympiad Google DeepMind · 21 July 2025
“An advanced version of Gemini Deep Think solved five out of the six IMO problems perfectly, earning 35 total points, and achieving gold-medal level performance.”
-
E2 AIME 2025 AI Wiki (aiwiki.ai)
“| Gemini 3 Pro | 95.0 | pass@1 (no tools) | Google DeepMind | 2025-11[^17] |”
-
E3 FrontierMath Tiers 1-3 (v1, old) Epoch AI
“| gemini-3-pro-preview | FrontierMath-2025-02-28-Private | 38% (±3%) | 10/290=3% |”
-
E4 AIME 2025 | AI Benchmarks | Longterm Wiki Longterm Wiki · 14 October 2026
“| 4 | o3-mini | OpenAI | 86.5% |”
Reasoning
Before the judgement date, an advanced Gemini Deep Think model scored at gold-medal level at the 2025 International Mathematical Olympiad, solving five of six problems (E1). Benchmark tables show more than one model scoring very high on AIME 2025: Gemini 3 Pro at 95.0% pass@1 without tools (E2), and o3-mini ranked only fourth at 86.5%, which means at least three models scored higher (E4). On the FrontierMath Tiers 1-3 benchmark, Gemini 3 Pro scored 38% (E3). The material does not say how difficult this benchmark is or how this score compares with other models, so it does not on its own show strength or weakness in a STEM area. The evidence covers mainly maths, but it shows several leading models performing at a high level on STEM tasks, which meets the criteria for true.
Cite this claim
Daniela Amodei, “A lot of current AI models are very good at STEM.” (said 7 February 2026). Verdict: resolved true. Reliability API, claim c_3f72af86782f, method 0.1.0+28da144b, dataset release v1.1.2. https://epistacklabs.com/claims/c_3f72af86782f
This claim is in dataset release v1.1.2, with the same verdict.
Reply to this claim
More factual claims by Daniela Amodei
-
Mostly true
“In the lab, Claude designed new protein binders from scratch.”
In laboratory work, Anthropic's Claude designed new protein binders from scratch.
-
Mostly true
“including open-source code much of the internet runs on, before attackers could exploit them.”
-
Resolved true
“They also showed that more than 6,000 women had come in for prenatal care, but only 1,762 had delivered with a healthcare worker present, a discrepancy that the team is investigating and addressing.”
-
Resolved true
“Those dashboards flagged a spike of 25,000 pneumonia cases, so the team can anticipate and plan for future surges.”
-
Resolved true
“Sometimes Claude is wrong.”
Verdicts are made automatically; how verdicts are made.