Skepthical Study - Machine Learning False Positive Flags: leaderboard
Skepthical tests models and review pipelines on deliberately flawed research papers. Astrophysics and machine-learning results use separate 15-paper cohorts and expert error annotations. Free-form, structured and concise review prompts remain distinct. False-positive rates use all raised flags on the clean-paper controls. This board reports clearly false flags / all raised flags on clean copies (%). Lower is better.
Metric: Clearly false flags / all raised flags on clean copies (%). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 3 models tracked. Updated 14 Nov 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Gemini 3.1 Pro Preview | Evaluation run: Skepthical Study Free Form | 3.5 | |
| 2 | Claude Opus 4.7 | Evaluation run: Skepthical Study Free Form | 5.9 | #78 |
| 3 | GPT-5.5 | Evaluation run: Skepthical Study Free Form | 9.2 | #52 |
Interactive version: theaggregate.ai/benchmark?slug=skepthical-study-machine-learning-false-positive-flags · How It Works · Data refreshed daily, snapshot 2026-11-16.