A new field of research aims to address concerns about the growing use of AI in medicine. But so-called explainable artificial intelligence may not help and, in some instances, can hurt.
