Tristan Harris

So exactly.

So we might not even have had that option.

But that's why it's like, I invoke that because it's like, that's something that no one wants to say.

And I'm not saying that to fear people.

I'm saying that to say, if we don't want to have to take that kind of extreme action, relative to that extreme action, there's much more reasonable things we can do right now.

Mm-hmm.

We can pass laws.

We can have the Vatican make an interfaith statement saying, we don't want super intelligent gods that are created by people who don't believe in God.

We can have countries come to the table and say, just like we did for nuclear nonproliferation, we can regulate the global supply of compute in the world and know we're monitoring and enforcement all of the computers.

What uranium was for nuclear weapons, all these advanced GPUs are for building this really crazy technology.

And if we could build a monitoring and verification infrastructure for that, which is hard, and there's people working on that every day, you can have zero knowledge proofs that have people say limited, semi-confidential things about each other's clusters.

You can build agreements that would enable something else to be possible.

We cannot ship AI companions to kids that cause mass suicides.

We cannot build AI tutors that just cause mass attachment disorders.

We can do narrow tutors.

We can do narrow AIs.

We can have stronger whistleblower protections.

We can have liability laws that don't repeat the mistake of social media so that harms are actually on balance sheets.

That creates the incentive for more responsible innovation.

There's a hundred things that we could do.