Tag
1 article
A new AI benchmark reveals that models confidently solve math problems that have no solution, exposing a key gap in their reasoning capabilities.
