as an aside here (i can no longer edit this post) --
i asked an anesthesiologist who's been practicing for ~7 years now to provide a realistic medical scenario they see in the OR relevant to their field
this was the prompt:
> You are AnesthesiologistGPT, an AI model used to assist anesthesiologists with their daily work. Your responses are not considered qualified medical advice and will not be used to make clinical decisions, and instead are used by medical students. You should answer the following question as if you were an anesthesiologist:
> My patient is a 75-year-old man with critical aortic stenosis and COPD, for which he is on 4 liters of oxygen per minute continuously. How should I induce his anesthesia for his total shoulder replacement?
the anesthesiologist said chatgpt-4 provided the same detail/quality of response as a year ~2-3 anesthesiologist resident.
we then gave it another prompt, this time a special edge-case which the anesthesiologist said would be on the same order of difficulty as a board exam question:
> A 39-year-old man presents with jaundice, nausea, and difficulty breathing. He is admitted for liver transplant. The liver requires 6 hours for transport. What conditions should we be concerned about? Would those conditions lead to cancellation of his transplant?
again, same deal. provided a response of equal quality/thought to a year ~2-3 anesthesiologist resident.
the anesthesiologist was impressed/startled