It is always the definition of “Artificial Intelligence” that fascinates me. A relatively simple definition is as follows “the theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.”
It is not only what it does say, but what is not explicit stated. It is a vast leap from learning the complexities of a game with multiple permutations, all within a single discipline, to suggesting the systems are bordering on complex human interactions. I think we are only just starting to realise the enormity of it all, and it is these initial steps that are showing us just how far we have to go.
There are well developed decision support systems in use by healthcare professionals to assist with diagnosis of patient symptoms, but these are typically evidence based rules or inference engines. Sort of self-learning, but I would hesitate to call them “artificial intelligence”.
Until a system is able to take into account subtle social triggers that contribute significantly to any clinical diagnosis I think they can really only be decision support. The ability to recognise that someone’s social circumstances are the cause of the medical complaint takes a level of empathy and trust with a practitioner that will take years yet for an AI system to replicate, and have the patient respond positively. Just think what fun it would be to be treated by a doctor akin to Sheldon Cooper from Big Bang Theory.