SPEAKER_04

And there's no expert for any of that.

But AI can be, at least in part, that expert.

So suddenly I have 22 million papers published today.

In all the fields of science, several tens of millions just in immunology alone, an AI can be the sleuth for me, can be both the angel and the devil on my shoulder that can make sense of things in ways that I never would have been able to before.

Especially with agentic AI.

So we, for instance, in my lab have developed an agentic AI that is basically an immunologist scientist in a box.

We can give it the raw data and we can pose a question in natural language.

And then we say, hey, make sense of this and turn it into a network.

Normally, that would have taken a graduate student along with a couple of postdocs months and months and months to put it all together.

Now, in three hours, we can get pictures and hypotheses of how all that data fits together in ways that I never could have done before.

In the beginning, it did a lot of hallucinations, which you probably heard about in AI.

But my answer to my colleagues is some of my best students hallucinate.

Right.

Right.

And and so but, you know, the human still in the loop.

And so with all of this together, now we can make meaning out of the data and we can skip a lot of the intermediary steps and speed it up.

And it's just getting better.

I mean, we, for instance, have put in a couple of papers now where.

So, for instance, in where my special one of my recent specialties is what's called the tumor immune interface.

So you have the tumor, you have the immune system, which is coalescing on near.