Whatever is the case today, within a few years the automated systems will clearly outperform doctors on most cases. Today we have systems that can hallucinate r

Whatever is the case today, within a few years the automated systems will clearly outperform doctors on most cases. Today we have systems that can hallucinate randomly. Since most of it is a black box, we cannot tell when it is hallucinating. This is solvable and just a matter of time. Today we do not have feedback to the automated systems. If we start doing following on a large scale Measure->treatment->measure again->adapt treatment, then the system will learn and will make connections that no doctor has ever made, because the minds of all doctors are not interconnected, at least not in a structural manner.