NB: So there were two questions there. Let me try to say something on both. A lot of what I've written about is how things could go wrong and existential risks and stuff like that. So in some sense, it's sad that it still looks plausible. It would've been better if that had all turned out to be false and barking up the wrong tree because then the world would be safer.
The fact that now as we move closer with AI, there do really seem to be these big difficulties with alignment and so on, is an unfortunate vindication.
As for the second question, I think to some degree it increases the credence somebody should assign to the simulation hypothesis if they are in a slot that would be more likely to be disproportionately frequently simulated. I don't think it's a huge effect relative to what everybody has reason to believe.
You could take some even more extreme case, perhaps, if you were Donald Trump or Elon Musk. If you’re reflecting, you got to at some point wonder what are the chances that I would just happen to be Donald Trump or Elon Musk.
MR: There’s all this debate about the traditional anthropic principle vis-à-vis humans, but you’ve coined a real anthropic principle.
NB: It is. It’s also the same with the simulation argument. If you're the person who published it, it also creates some slightly higher salience. If you imagine simulating not everybody, but just some people.
MR: Does your gut say that I’m a real person sitting here asking you these questions?
NB: Yeah, it does. And in fact, even if this is a simulation, I still think that you and I are real in the sense that matters — that we would be having real experiences, our actions would matter, and so on. It's just that the nature of what that reality consists in would have the surprising property of being implemented in some computer built by an advanced civilization.