Quanquan Gu: not convinced AI labs should stop treating math problems as benchmarks

QuanquanGu · x · 2026-09-13

Quanquan Gu narrows his disagreement with Csaba Szepesvari: he is not fully convinced AI companies should be asked to stop treating important mathematical problems as benchmarks. Szepesvari notes the letter's only request was precisely that — not a claim that AI is bad, as it praises its potential — but that it isn't working.

Related event: Szepesvári Defends Mathematicians' Open Letter: Target Is AI Benchmark Culture, Not AI Itself(5 posts)→

Original post →

More from AGI Musings

AGI Musings channel →