Study: Chat GPT is better than doctors at diagnosing illness

A new research study indicates that Open AI’s chatbot Chat GPT-4 is better at diagnosing diseases than human doctors, according to The New York Times.

Fifty doctors, a mix of attending physicians and residents, participated in the study; diagnoses were based on evaluations of medical patient cases. All in all, Chat GPT-4 got a 90% score for the diagnoses it delivered; the doctors on their own got average scores of 74%.

The doctors also reportedly performed worse than the chatbot when they were allowed to use Chat GPT-4 in their work. Physicians who used the tool performed only marginally better — getting scores of 76% — than physicians who did not use a chatbot at all.

The reason for that, the study said, was that the doctors rarely allowed themselves to be convinced by the chatbot if it noticed something that did not agree with the doctor’s own diagnosis. Many doctors also didn’t know how to use the chatbot’s skills to their full potential.