Toby Ord: RSI scaling may mirror pretraining — slow error gains, huge revenue payoff
tobyordoxford · x · 2026-10-21
Researcher Toby Ord joined a thread on measuring recursive self-improvement, arguing what people want to estimate is essentially β from the RSI literature: if efficiency (A) scales logarithmically with CoT, β grows without bound as A increases, forcing r toward zero.
He drew an analogy to pretraining scaling: Kaplan and Chinchilla scaling laws showed error improving painfully slowly with compute (around compute^1/20), yet reality contained an impressive function transforming error into revenue (error^-20) — meaning modest capability gains can still translate into massive economic value.
More from AGI Musings
- When AI saves you an hour at work, who decides where that hour goes? — Alchemistwiza · 2026-10-21
- Quantum+AI researcher asks: are we optimizing for the right goals? — AryHHAry · 2026-10-21
- Nadella: SaaS Will Collapse in the Agent Era — But One Rebuild Cost More Than 2 Years of Subscription — msharmas · 2026-10-21
- AI May Soon Make Scientific Breakthroughs Nobody Can Explain, warns AI commentator — VraserX · 2026-10-21
- Mathematicians clash: do self-maintaining biological neurons make AI fundamentally different? — ValerioCapraro · 2026-10-21
- "Every job was a placeholder until a machine could do it" — bold AGI labor take — rand_longevity · 2026-10-21