Ethan Mollick marvels that next-token prediction got AI this far

emollick · x · 2026-11-21

Ethan Mollick reflects on how unreasonably effective human language has proven at modeling the universe — simply training a system to predict the next word in a sentence got us remarkably far. A concise observation on the surprising power of next-token prediction as the foundation of modern AI.

Related event: Ethan Mollick: LLM Success Is a Win for the Humanities(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →