I have two minds about this. GPT talks with strong confidence about things, and often in ways where it’s not immediately obvious what is incorrect. That can be

I have two minds about this. GPT talks with strong confidence about things, and often in ways where it’s not immediately obvious what is incorrect. That can be a bad thing if the user is unaware of it. Certainly don’t mainline advice from a statistical model into your health practices. On the other hand, this could be an extremely powerful tool for exploration. If it is used to come up with possible issues for an expert to disprove, it could be helpful, depending on the signal to noise ratio. Complicating this is the obscene costs of the American healthcare system, which I worry makes the human part of a potential human-in-the-loop system out of reach for many people.