Comment by chezelenkoooo

The author touches on this in the very article linked. His point is that the low accuracy answers provided by ChatGPT risks reinforcing the doctor's biases and then not thinking outside the box. In other words it risks the doctor blindly trusting the AI too much.