METR's famous time-horizon chart may measure something else — latent capability fits with R²=0.996
wfithian · x · 2026-10-16
Researchers improved the estimates behind METR's famous AI time-horizon chart, and the curve visibly changes. But converting both estimates to a latent capability scale recovers near-perfect agreement (R² = 0.996), suggesting the time-horizon trend may largely reflect underlying general capability growth rather than pure task-duration measurement — an important methodological caveat for anyone extrapolating AI timelines from that chart.
More from Research
- Researchers derive a dropout scaling law from active parameter count — gordic_aleksa · 2026-10-16
- Gemini is a robotics outlier: dropping it lifts capability-robotics correlation from 0.20 to 0.96 — ycombinator · 2026-10-16
- EPFL team uses RFdiffusion to design de novo metamorphic proteins switching between two folds — ferruz_noelia · 2026-10-16
- Manipulating a model's sense of power may reveal its takeover counterfactuals — jd_pressman · 2026-10-16
- Poser benchmark unmasks alignment-faking LLMs, one detection method catches 98% — jd_pressman · 2026-10-16
- Six Years of Evidence That MLP Neurons Are an Interpretable Basis, in One Thread — aryaman2020 · 2026-10-16