Rob, Luisa, and the 80000 Hours team

I guess that's...

There's a sense in which that's obviously really good, but then maybe it should make us queasy, the idea of a government going in and basically writing what assistance you can get with anything.

Okay, so that's why AI constitutions are going to be important.

I guess, is there anything to be done on this today?

Yeah, I guess... So the way that we've tried to safeguard at the written constitution stage is making it very difficult to change.

Or I guess you bring everyone together and then try to agree a constitution that a supermajority is in favor of, and then you make it really difficult to change it.

I guess it's a very difficult balance to say how difficult is the right amount of difficult to change.

But we're not going to be able to do that with AI, right?

Because it's changing so quickly, and they're always ordering the system prompt, or at least it doesn't seem...

You know, if we passed a law saying the system for everything has to be exactly this and we can't change it.

You're really making me love open source AI, David.

I think I never felt so enthusiastic about it as this minute.

Yeah, it's a difficult challenge because it also creates some problems.

Do you think we'll navigate that one reasonably?

Okay, the open source versus non-open source thing?

I guess you expect excessive government control?

Okay, and so you're less worried about the bioweapons stuff?

Yeah.

I suppose the idea will be, again, a keyhole solution that fixes the specific problems with open source AI, the ways in which it's most dangerous and lets the other problems slide and then we use it as much as possible.

So there was another essay earlier this year, I think, The Intelligence Curse by Luke Drago and was it Rolf Lane?