OpenAI's o1 reasoning model outperforms physicians at diagnosis in Harvard-led Science study

unverified 09 May, 16:45

OpenAI's o1 reasoning model outperforms physicians at diagnosis in Harvard-led Science study

OpenAI's o1 model diagnosed 76 emergency room cases in Boston with 67% accuracy, outperforming two expert physicians (55% and 50%). Reviewers could not distinguish AI notes from doctors'. Published in Science.

#science

Published on

TikTokYouTubeThreadsBlueskyFacebookInstagram