Eli Lifland: lab 'training pauses' likely oversold, only ~25% chance of real slowdown

eli_lifland · x · 2026-09-30

Forecaster Eli Lifland is skeptical that a frontier lab's claimed training pause means a real slowdown: he puts 25% on a substantial slowdown, arguing paused compute likely shifts to experiments with little impact on months-scale progress, and that public statements tend to oversell the deceleration.

Related event: Forecaster Eli Lifland doubts training pause will meaningfully slow frontier AI, puts ~25% odds on real slowdown(6 posts)→

Original post →

More from AGI Musings

AGI Musings channel →