Then I'm Like, Hey, Fergal, I Want Your Team to Use This for Medical Diagnosis?

Fergal: "It's not going to like, at least my experience of it directly and even what you said about say the two FA use cases" Fergal says there is a very low probability of accuracy. But he adds that while it will still make things up, it's getting better and better at also giving you the right answer when it has the right answer. The tech could be used for medical diagnosis or as part of a legal case but would need some kind of cladding around it before being deployed in production settings.