Doctors are given cases to diagnose, part of which gets access to GPT-4 to help. The control group scored 73% right and the GPT-4 group 77%. There is no big difference.
But GPT-4 alone scored 92%. Doctors didn’t want to listen to AI.
Learn more about the host, Ethan Mollick.
The post Model this appeared first on Marginal REVOLUTION.
Source link