AI scores a ‘C-’ on its hardest math test yet

Scientific American Scientific American

The second batch of “First Proof” problems is meant to evaluate AI’s usefulness for research-level math.

The best model got six or seven of the ten questions right.

Read full article at Scientific American →