A Harvard Medical School-led study found that an OpenAI large language model outperformed human doctors in diagnosing emergency room patients, with higher accuracy in both limited and detailed data ...