So what are some of the things you do to make sure that AI is being deployed in a thoughtful, safe, effective way? Assuming we can't keep AI from hallucinating completely, how do we sort of mitigate the dangers of the hallucination?

Right, so let's start with that hallucination question first.

IQVIA has taken a very particular approach to this, which is we do everything through prompt-driven activity.

So we start off by limiting the scope by which something can hallucinate because what we're looking for, I think we have enough evidence that if you write the prompt in the right way, you'll get the right answer.

or that the AI is capable of giving you a good answer.

So we are very focused on limiting the size of any prompt to avoid hallucinations.

I'm not saying they never happen, but we've taken that approach to basically mitigate that as much as practically possible.

So how do you make sure that as you're working with organizations, that they're You know, that everyone from the top of the C-suite down to the, you know, most entry level employees who are working on this is, you know, kind of knows what's going on with the AI and is, you know, on board with it and is working, you know, in the right way.