A new study reveals that human mathematicians have surpassed AI in solving unpublished high-level math problems, challenging ...
The second batch of “First Proof” problems is meant to evaluate AI’s usefulness for research-level math. The best model got ...