AI health care tools appear to have a reliability problem, and experts suggest a new way of evaluating them: Study their performance together with the humans who use them.
Why it matters: Hundreds of health care AI companies raised millions at sky-high valuations for AI tools on the premise of high accuracy.
