Testing AI systems on hard math problems shows they still perform very poorly
As developers of AI systems work to improve the math skills of their models, they have developed benchmarks to serve as a ...