A reflection on the LLM debate and AI epistemology

antirez · x · 2026-07-20

A short reflection on the long-running LLM debate: the author argues that some critical thinking was enough to realize these systems were never just simple next-token guessers, and says the episode will end up in the history of how we understand knowledge.

The reply adds that recent results also undermine another mistaken belief — that many AI scientists who were skeptics about LLMs were right — and notes that even experts in the field can make obvious errors because AI is still not fully understood.

Related event: Community Reflects on AI Epistemology and Inadequate Evaluations(3 posts)→

Original post →

More from AGI Musings

AGI Musings channel →