Terence Tao: LLM-math skeptic's only 'experiment' was using Copilot

chaumian · x · 2026-09-21

Terence Tao shares an anecdote: at a lunch in late 2024 he argued LLMs could do decent mathematical reasoning, but rperezmarco wasn't buying it. When asked which models he had actually experimented with, the skeptic's answer was "Copilot." A pointed jab at critics dismissing LLM capabilities without firsthand testing.

Original post →

More from AGI Musings

AGI Musings channel →