Grant Harvey

And with the Charlotte Hornets, they were thinking, well, like, it'd be really cool if we could just watch every game of basketball in the US to assess players.

And obviously, I mean, it's probably an inefficient use of their funds to hire 100,000 scouts.

But now with, I mean, maybe I've just revolutionized basketball.

Maybe they'll be like, this is the move.

We just hire 100,000 scouts.

But I doubt it.

But instead we were like, okay, well, we can get a video camera in these places.

And actually computer vision models are now much more advanced that we can start to pull out way more relevant data from these games.

And we can start to pull out patterns and we can start to pull out sort of recognitions about the movements people make or the shapes people form.

And suddenly you've effectively got a scout in every game.

Exactly.

Because basically, historically, and this is why sabermetrics was such an interesting thing was because

Baseball is a very statistics-heavy game because lots of these things are measurable.

But now, things that were historically viewed as unmeasurable because they were unstructured data, you can actually analyze more effectively.

I mean, that's what language models and other models do.

They take unstructured data and do interesting potential stuff with it.

Not guaranteed.

It could be complete dross.

But they give you the ability to interrogate it more effectively.

Just think conceptually, a large language model