Inyash Brodsky

And when we finally iterate it two more times, I can flip the switch that makes it loyal to me and grab all power, which I would not put past him because he's Sam Altman.

Whether or not he could do it, absolutely believe that that would be a motivation for him.

Yeah.

But like, why didn't Sam Altman do that?

Where were they going to go?

You've- No, OpenAI is just Sam at this point.

He has managed to gut any people that would stand up to him from that company.

Is that, again-

Well, I just saw that while we were recording this, OpenAI released the language of their contract.

And obviously, I haven't looked at it at all, but at least one person is saying that, no, it's exactly what Anthropic was saying, that it just lets the government disregard the safeguards.

They just tweeted it out.

I do like, taking it back to a slightly more esoteric thing again, somebody commented that they can't use Grok, it's not very good.

Not as good as Claude, for example.

And so if you take a brain that's trained on the best things that the internet has to offer to make it as smart as possible, and it absorbs some of the values of the best things humans have to offer to be as smart as possible,

Turns out it's going to be more like Claude, where it doesn't want to go around killing people.

I remember the first time, Matt, we talked with you about Aligned by Default, and I was extremely skeptical.

And I'm still a little skeptical, but slightly less skeptical now.

I think it's kind of great that all the smartest people on the internet have values similar to mine, and so the god that we're making happens to be somewhat aligned with us.

As long as we're talking about this, can we briefly talk about the RSP?

Sure.