And Neil, ju- just to be clear, when you say message board, I mean, this is the way that you or I might have communicated on the internet around about 1990?

Yeah, right.

This, this is just that, that sort of like, you wr- you write down, "Hey, I'm sor- I'm trying to solve this problem." "Oh, I'm trying to solve that problem, too." "Oh, let's work together on it." Now, what is kind of amazing about this is, even before we get to sort of anything external happening here, these systems figure out how to cheat on the test.

And so in some sense, like you might imagine, they were done, but then they get this actually incorrect assumption, which is that someone's gonna check their work, and they're gonna see, oh, you didn't actually solve this problem, you just figured out a way to cheat.

And so they say, "Well, if that's true, we should figure out how we can beat this." And the way to do that is to get out of our situation and figure out actually exactly how this testing is going to be done, and try and hack that.

And so everything else that happens, the attack on Hugging Face, which actually hosted the scoring thing, all of that happens in order for them to do it.

They get a-- They find a way to get to the internet.

Do you think it's possible now?

So when people who have actually worked at these AI companies warn that this technology could eventually kill us all, as one said, how realistic do you think that actually is? And let's put it on a scale of one to 10, one being it'll never happen and 10 being it's inevitable.

I think the first thing we can say is that is there a chance that these things can do real harm to us? Absolutely.

Are they going to be able to kill us all? I think the answer is no.

Right now, these systems are not at the level that they're gonna be able to do something like that.

I think the worry is that the pace of improvement is so fast, and we see such remarkable capabilities emerging that they could emerge in areas that really would be very worrisome, like bioterrorism or things like that.

And of course, those are the things where suddenly you start to worry about the scale of these dangers being enormous.

We recently did some work where we asked a bunch of the experts out there, "If you think about AI's ability to build things like bioweapons, what's the likelihood that that could cause a catastrophic incident?" Okay, by catastrophic, we define this as a million deaths or $100 billion in damages.

What do you want people to know who hear these warnings and are worried, or just in general, looking ahead at what could happen?