METR: AI acceleration trend findings swing with methodology choices
CFGeek · x · 2026-11-21
METR's trend analysis of its index finds anything from slight deceleration to moderate acceleration depending on methodological choices, using only public models through September and thus not directly comparable to Anthropic's "2x acceleration on AECI" result. Appendix H details comparisons: MCI slope changes broadly match METR's October 2026 AI R&D Acceleration report under similar calibration, though usually lower than Anthropic's system-card figures; linked linear fits to IRT models are sensitive to calibration models, breakpoints, and benchmark choice. On public models up to Opus 5.5 and GPT-6 Astra, positive slope changes appear within Anthropic models but not OpenAI's.
More from Research
- Shendure lab preprint uses CRISPRa to screen chromatin-opening transcription factors at scale — anshulkundaje · 2026-11-21
- Guava distills GPT-5.4 into a 4B robot agent using 2,000 sim trajectories, zero-shot on real hardware — ChongZzZhang · 2026-11-21
- Facet-SLAM Hits 113 FPS on RTX 4090 with Triangle Splatting Real-Time SLAM — jonstephens85 · 2026-11-21
- Slay the Spire joins AI eval suites; researcher notes only Ironclad was tested — najoungkim · 2026-11-21
- RH proof author Masullo: a human came up with these ideas, I'll see it through — Singularitarian · 2026-11-21
- DPP visualized: selection probability scales with squared spanned volume — xtimv · 2026-11-21