Statistical diagnoses models have offered similar possibilities in medicine for 50 years. Pretty much, the idea is that you can get a far more accurate diagnosis if you take into account the medical history of everyone else in your family, town, workplace, residence and put all of it into a big statistical model, on top of your symptoms and history.

However, medical secrecy, processes and laws prevent such things, even if they would save lives.

I don't see ChatGPT being any different.