Optimistic Prediction Under Frontier Model Regulation

scaling01 · x · 2026-07-11

The author states that he makes these predictions not to "get it right," but to bet on a slightly optimistic accelerationist outcome: his estimates are deliberately on the high side, not median judgments. He also mentions that compared to his confidence in ARC-AGI-2 last year, he is less sure this time, partly because **frontier labs/models face stronger government control**. However, he still thinks **170 is feasible**.

Original post →

More from AGI Musings

AGI Musings channel →