Tristan Harris

speaker

532 appearances 2 recordings 1 series first heard Aug 2026 last heard 16 Sep

Tristan Harris’s voice in public audio — every appearance, attributed to the second.

Trend

recordings per month · last 12 months

Recordings per month over the last 12 months — 2 in all, peaking in Sep 2026 with 1.

newest first · ▶ plays the moment

Yes, I think that's right.

And you're making a really important point, which is that the idea that regulating ourselves at all just means that we'll lose to China.

I think this is so misguided because, for example, the US beat China to the technology of social media, but we didn't govern it very well.

So we invented this psychological bazooka, and then we didn't actually regulate it.

So it had the bad incentives, and we spun the thing around and blew off our own brain with it.

So did beating China to social media make us stronger, out-compete them, or did it make us weaker?

Well, it made us weaker.

Well, what you're pointing to, I think, is that a lot of the Chinese AI progress has been based on copying and distilling the U.S.

model.

So we say we have to race to beat China, but there's a joke, you know, the GIFs online, these memes.

There's a meme where there's a guy in a speedboat, that's the U.S.

and the frontier models.

And there's a guy behind him on some water skis, like with attached with a little rope, right?

Basically, the guy in the water skis is the Chinese AI company that's tied to the speedboat.

The water ski guy is going, go faster, go faster, because the faster you go, the faster I catch up to you.

I'm always right behind you because I'm either inside your companies, I'm stealing your AI models, I'm hacking your AI models and getting them

There's people who believe that that happened very quickly after mythos was released and China got access to the models.

And then we make progress and then they just distill the models.

It basically means like, I'll take your digital brain, I'll like run it a million times and get a bunch of outputs and then train on its outputs.

And then I'll basically catch up to where you are.

Showing 421–440 of 532 · page 22 of 27