Toby Ord: correlated benchmark-ceiling errors weaken evidence for AECI slope increase
tobyordoxford · x · 2026-11-08
Toby Ord joins the debate over METR's capability growth estimates, arguing the evidence for any slope increase is weak: while the central points do suggest an increasing slope, they sit at the upper range of available benchmarks and carry correlated errors, which may overstate the trend. His comments respond to Jacques Thibs's claim that METR's fit missed a joint jump-plus-slope-increase model.
More from AGI Musings
- NY Assemblyman urges Congress to ban Recursive Self-Improvement this year, citing 2027-28 AI lab timelines — rohanpaul_ai · 2026-11-08
- Radziwill says the quasi-Riemann proof fits in a tweet from the future — Singularitarian · 2026-11-08
- DeepMind exec says scientific discovery is undergoing a 'genuine phase transition' in pace — pushmeet · 2026-11-08
- IDC analyst: AI claims on social media distort completely after two reshares — perilli · 2026-11-08
- AI progress per dollar fell 63% as R&D spending grew 5.3x faster than capability gains — kimmonismus · 2026-11-08
- Perplexity CEO on Joe Rogan: the desk job was a sales strategy, and AI taking it over isn't bad — 2C_ornot2C · 2026-11-08