Mathematicians we're told to hunt the 'least satisfying' AI proofs

RexDouglass · x · 2026-09-14

A debate on AI in math research: brute-forced AI proofs keep landing, but mathematicians find them aesthetically unsatisfying — like writing assembly when Rust was promised.

jonmellon proposes a heuristic: pursue exactly the solutions the math community would find least satisfying, since that's where AI genuinely advances the frontier. RexDouglass pushes back on claims that elegant new math hides behind every brute-force result, 'just 25 years away' from human discovery.

Related event: Mathematicians Debate Whether AI's Brute-Force Proofs Are Elegant(3 posts)→

Original post →

More from AGI Musings

AGI Musings channel →