Toby Ord puts ~1-in-3 odds on dramatic AI takeoff from recursive self-improvement by 2030

tobyordoxford · x · 2026-09-14

Oxford senior research fellow Toby Ord offered his take on Anthropic's claims around recursive self-improvement: he thinks RSI probably won't drive a dramatic takeoff of AI capabilities by 2030, but puts a decent chance on it — around 1 in 3.

His key point: even if the likely outcome is no takeoff, the sheer stakes mean we should take that probability extremely seriously.

Related event: Oxford's Toby Ord puts odds of AI intelligence explosion before 2030 at one-third(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →