An unreleased Anthropic model made progress on one of math’s biggest unsolved problems
TL;DR
The Riemann hypothesis has stood as one of mathematics' major unsolved problems for more than 150 years. Anthropic has not solved it, but an unreleased model from the company made more progress than most people would expect. The result is not a proof, it is an incremental step on terrain where models have contributed very little so far.
Nauti's Take
A model making genuine headway on a 150 year old problem is real progress: AI is moving beyond text production into a role as a tool in basic research. The limit is still obvious though, a partial step is not a proof, and without human mathematical review a result like this is worthless.
The interesting move is for research teams to test models as idea generators, not as an authority.
Sources
