FrontierMath Erdős announced as a new AI math benchmark
The announcement describes a collaboration with @thomasfbloom aimed at improving the scientific rigor of evidence about AI math ability.
TLDR
The FrontierMath Erdős announcement cites Tao’s remarks at the ICM that evidence on AI math ability has mostly not been gathered under proper scientific conditions: “it's a confusing mess.” The author presents the new benchmark as an attempt to start changing that.
Combined views
—
1 Source, first seen 27d ago
likes