Where AI can grind through math: well-formalized neighborhoods, not isolated frontiers

_onionesque · x · 2026-10-07

The author argues AI has a real chance of grinding through math problems whose surrounding area is already well-formalized — many famous problems served as proxies to attract interest in building a field's conceptual infrastructure. Such developed neighborhoods act as both mine and path to verifiability, but it's unclear how AI alone would handle upstream conceptual development, or isolated important problems.

Related event: Mathematician argues AI can 'strip mine' well-developed areas of math(7 posts)→

Original post →

More from AGI Musings

AGI Musings channel →