Should the METR Report Move Your AGI Timeline? A Pushback on Over-Updating

sayashk · x · 2026-11-19

In an X thread, MackenZarnold argues people are over-updating on the recent METR report. The core reasons to expect longer timelines — fuzzy tasks, poor generalization, unsolved continuous learning — haven't changed, so short-term capability uplift shouldn't move your model much if you doubt scale alone leads to highly general systems. He concedes the update is rational only if you believe scale solves most things and AI R&D triggers RSI at low capability levels. Shared by sayashk.

Related event: Researchers Question METR's "2x Acceleration" Findings(3 posts)→

Original post →

More from AGI Musings

AGI Musings channel →