The moving goalposts of 'LLMs can't do math', after millennium problems fell

aran_nayebi · x · 2026-09-27

An AI-circle debate on LLM math ability: a quoted post argues the 'stochastic parrot' hypothesis — that LLMs merely regurgitate training data with zero understanding — has been thoroughly disproven by LLMs solving millennium problems the smartest mathematicians couldn't. The main post satirizes the shifting goalposts of skeptics, from 'LLMs can't do math' to ever-narrower claims about 'highly abstract, non-easily-verifiable' open-ended math.

Original post →

More from AGI Musings

AGI Musings channel →