Jonathan McRae

Because not all scientific research is possible to do in the cloud.

The first question I have to you about the methodology is the creative approach.

So when we think of scientific hypotheses, a researcher comes up with an idea that they want to test, how do they do that in the real world

And how do you get an AI to do that when it's not a human?

And is it good at that part of the process?

Because there is, you know, depending on who you ask, you know, creativity of AI can sometimes be quite limited when it comes to the applications of, you know, having read what's already there.

Is it good at coming up with hypotheses to test?

But the AI didn't just read that paper and present it as its own.

So let's talk about the underlying technology.

In terms of the local models that you're using, obviously you're training them yourself to some certain degree, but are you using your own LLMs or are you...

Are you accessing the commercial-sized LLMs like Opus 4.6 and Cloud, or the latest versions that are available commercially?

Yeah, and my other hat, I do a lot of AI literacy, and I think at this particular moment in time, there is a chasm of understanding between what is possible and what people think is possible.

I think a lot of people have not realized how good these models have gotten in the last three months.

Yeah, and I fully agree with that because, of course, one of the things that AI can do that humans can't, and I was at the launch of Charlotte Bleese's brilliant book, Dr. Bot, yesterday, where she talked about, you know, how AI might go around solving certain health problems.

But one of the things she said

she kind of mentioned, which kind of stood out to me, is the enormous amount of medical literature that doctors are expected to keep track of.

And she said, even if it was just 2% of the latest developments in the medical literature, doctors would need to be reading 25 hours a day.

And so one thing that an AI can do is take what is already known and look at other methods in other fields, other technologies, and be across those in lightning speed.

And that gives AI such an incredible

over humans in terms of joining up dots.