Anthropic's AI cracks decades-old 'holy grail' conjecture in probability theory

Dr_Singularity · x · 2026-10-01

Per Scientific American, Anthropic says its AI generated a proof resolving a famous conjecture in percolation theory — the study of when random networks become permeable — a problem long considered a 'holy grail' of the field. Mathematician Benedikt Jahnel had said solving it would plausibly merit a Fields Medal, and 2022 medalist Hugo Duminil-Copin himself failed. Days after Duminil-Copin predicted on Aug 30, 2026 that AI would eventually topple the field's most famous conjecture, Anthropic announced its AI had done so. If verified, it would mark a landmark for AI at the frontier of mathematical research.

Related event: Anthropic's AI cracks decades-old percolation theory conjecture(3 posts)→

Original post →

More from AGI Musings

AGI Musings channel →