Priya Alexander

How did, just going with the really urgent life and death stuff, how did ChatGPT Health perform there?

Which could be fatal for someone.

How about the other end of the spectrum, Ashwin?

So for these non-urgent clinical settings where there's not a life-threatening situation, how did it perform there?

Was it good at managing people or did it over-investigate them or tell them to go to emergency departments?

Does it do well anywhere?

Does it perform well in the intermediate clinical group at all?

I just wonder, this appears to be in your study, the suggestion that if you inserted lab details, if you put in objective values like blood pressure or oxygen saturations, that it appeared to improve the accuracy of the CHAT-GBT-Health triaging?

Yes, there is.

Yeah.

A huge safety failure.

So what do you mean by that though?

Do you mean actually questioning what it tells you?

So to push back on it a little bit, to test it a little bit.

Okay.

A gentle test.

Okay.

Dr. Ashwin Ramaswamy from Mount Sinai Hospital in New York City.

No, I'm going to say listen to the episode because I think Ashwin had some excellent points.

But Norman, there's pros and cons to everything.