Patterson: almost no one grasps how fast AI progress will accelerate

davidpattersonx · x · 2026-11-18

Quoting his own earlier prediction, davidpattersonx argues that extrapolating exponential trends is easy, but estimating how fast those trends accelerate is not. He had already factored acceleration into his ASI-by-end-of-year prediction, yet acceleration now seems even faster than he expected — "it should have been obvious," since superintelligence itself will drive massive acceleration.

In the main post he splits people into three tiers: most expect a flat line into the future; a few can plot improvement trends and extrapolate; almost no one understands how fast improvement will accelerate.

Original post →

More from AGI Musings

AGI Musings channel →