Ajeya Cotra raises her 2026 AI-agent capability forecast
Ajeya Cotra said Claude Opus 4.6's roughly 12-hour METR time horizon made her January forecast too conservative, and revised her year-end expectation to more than 100 hours on comparable software tasks while stressing wide uncertainty.
0 comments · 0 votes
Sign in to join the discussion →
No comments yet. Start the discussion.
Why it moved the index
This is a materially changed, dated forecast grounded in METR measurements and explicit model comparisons. Cotra reports benchmark saturation and a wide confidence interval, but the observed jump from Claude Opus 4.5 to 4.6 led her to expect substantially longer autonomous software tasks and potentially easier decomposition of large projects. It is evidence of a forecast revision, not proof that full AI-R&D automation occurred.
Assessment history
- R1Toward 46 · confidence 67
Adds a source-backed forecast revision tied to exact models and measured agent-task horizons.
14 Aug 2026