Don't anthropomorphize LLMs: talk data coverage and error rates instead

gerardsans · x · 2026-11-21

Replying to a discussion on humanizing AI, Gerard Sans argues LLMs can't be smart, intelligent, or understanding in any meaningful sense — the more accurate framing is data coverage, sampling, utility, and error rates.

Original post →

More from AGI Musings

AGI Musings channel →