Toby Ord Questions AECI Acceleration Paper: Training Compute Not Ruled Out
tobyordoxford · x · 2026-11-07
Oxford philosopher Toby Ord discusses a paper claiming acceleration in AI research progress (measured in AECI). The thread notes the authors confidently exclude training-compute acceleration as the cause, but the evidence behind that conclusion is redacted, leaving readers to decide how much to defer. Ord says he couldn't interpret one claim, and his best guess is a large new pre-train boosting results by 5 AECI plus a 1.4x speed improvement—an explanation he believes the paper does not actually rule out.
More from Research
- Near-exact cache matches cause 2-7x more wrong replies than the verified gray zone — Reasonable_Royal_621 · 2026-11-07
- RLAF: Nace AI trains decision models with a reasoning model as judge — nischay_twt · 2026-11-07
- ICLR 2027 reviews criticized as unusually shallow: generic bullets, low scores without justification — Charming-Fail-772 · 2026-11-07
- Do as I Do: Human video as robot data — 'most robot data will soon be made without robots' — carlosdponx · 2026-11-07
- AI agents design and operate robot-powered labs, Nature news reports — AnnaCiaunica · 2026-11-07
- Toby Ord pushes back on METR's claim that AI acceleration isn't driven by training compute — tobyordoxford · 2026-11-07