Tristan Harris

So let's just stop there for a second.

So we're talking about compute.

By compute, we mean chips.

And then you're telling me that I think it would be a surprise to most listeners that the existing chips that are shipping out there actually have some kind of controls on them.

It would be like if we're shipping uranium around the world, but then the uranium says, well, if I'm being used for a nuclear power plant, I'll emit this signal.

And if I'm being used for a nuclear weapon, I'll emit this signal.

Now, this would strike people by surprise because they think a chip is a chip.

It just runs

you know, computation.

So help people understand, has this happened for a long time?

When did people put in this system?

Because what you're basically pointing to is there could be an optimistic case with AI, there is this finite resource of chips.

And so in this bottleneck, you're saying there's actually a way that that bottleneck could be controllable so that it could, for example, serve for an international agreement.

Could you explain what is telemetry?

So that would be like the electrical signals coming off of a chip or what kind of signals would we be talking about?

So we're talking about basically signals that you could pick up through those mechanisms that would tell you when you say training versus inference, just to remind listeners the difference between a chip that knows that it's training GPT-6 versus a chip that says, I'm only running GPT-5.

And we might have an international agreement where like everyone's allowed to run GPT-5, but you're not allowed to train GPT-6 because that would create this risk of some dangerous AI that we don't want to create.

Yeah.

And so you're saying that level of difference in the chip architecture would help us do that?

Right, so training in general versus we don't know what you're training, but theoretically that does point to something.