OpenAI thinks AI benchmarks are broken. Now the company is launching a program to fix how AI models are scored. The new OpenAI Pioneers Program will focus on creating evaluations for AI models ...