Jeff Klune

In fact, Newton invented calculus at the same time as Leibniz, and Darwin invented evolution at the same time as Wallace.

And this happens over and over and over again.

And in this case, the AI system looks at the previous discoveries.

It's been trained on the entire internet.

That's how things like ChatGPT work.

And then you can ask the system, hey, what would be an interesting new thing to study in machine learning?

What would be a way to improve the system?

So I would say that right now the system is pretty good.

It definitely comes up with creative, interesting ideas sometimes.

Not all of its ideas are great.

It's kind of at the level of an early graduate student in this regard.

But I actually think it's a bit of a distraction to focus on its current capabilities.

instead of the trend lines of how quickly it's getting better.

So there's a clear trend in machine learning, which is once something barely starts to work in a few years, it's going to be jaw-droppingly good.

And we've now kind of passed the Rubicon where it's barely starting to be good and soon it will be amazing.

So there are a couple of examples.

I mean, many people kind of do not like the idea

that AI scientists can be as creative as human scientists.

Humans really love to think that we're special.

We didn't want to not realize, we didn't want to accept that we weren't the center of the universe.