US Is Driving the AI Safety Race Off a Cliff | Nathan Labenz

Nathan Labenz (0:00)

It is our companies who have said that we are gonna do this recursive self-improvement plan. And by the way, we just hit the intern milestone and we're about 18 months from the fully automated AI researcher and hopefully it's aligned enough to make the next ones aligned. That's kind of the current state of play. That's coming from us, it's not really coming from them. And so I think we kind of need to look ourselves in the mirror when it comes to this idea that we can never make a deal with China. I'm afraid and it brings me no joy to say this because I'm a patriotic American like everybody else, I'm sure that you have on the show here. But I think unfortunately we're the ones right now that are making it feel so difficult to do a deal. And I think if we kind of free our minds and are willing to dream a little bit more about what a positive future, what a better relationship could look like, I think a whole new world of possibility might open up for us.

SPEAKER_2 (0:57)

All right, we are back. We're live with Nathan Labenz, who's the host of the Cognitive Revolution podcast. He's been writing about AI for a long time, and his podcast on is one of the best. So Nathan, it's great to have you on. Welcome to MTS.

Nathan Labenz (1:11)

Thank you. It's great to be here, and that's very kind.

SPEAKER_2 (1:14)

Absolutely. So what should we start with? What should we start with? You went to China recently and learned a lot about the Chinese AI ecosystem there. What were some of your top takeaways?

Nathan Labenz (1:29)

Yeah, it's a big topic. China, big country, big topic.

A joke that I learned when I was there, apparently originates with the foreign press corps years ago, is that somebody goes to China for a couple of weeks, they think they know a lot, they stay for a couple of months, they realize they only know a little, and if they stay for a couple of years, they'll realize they know nothing. So I'm very mindful of the limitations, certainly of my knowledge, but I was motivated to go because we're in this crazy AI moment. I'm not a China guy, really, I'm an AI guy, but we're in this crazy AI moment where just look at the news of the last 24 hours, we've got these loud resignations and people saying AI could kill us all, and no, we really do believe that. It's not marketing hype, seriously, we really do. And by the way, we don't have a plan. It's sounding more and more like cries for help from the frontier companies this summer.

And yet, for the last couple of years, any discussion it seemed like that I got into around, well, what can we do about this? You know, can we come together in some way to have these companies, you know, try to take a more perhaps collaborative or just safety conscious approach so they're not racing each other off the cliff? Those conversations would very often end in, well, China, you know, that sounds good, but you can never do a deal with China. They're never going to slow down. We, you know, do anything. We'll just be handing the race to China. And so that never really sat right with me, to be honest, but I was motivated to go because I figured at some point, I got to put my time and money where my mouth is and actually go see what's going on there.

I would say a couple of big things really stood out to me. One, China is, in my view, the most advanced futuristic feeling place that I have ever been.

So I think that's one thing we should really take to heart is like they are absolutely on our level. They are absolutely a peer global world power, technological leader.

And it's important to realize too that they've been there for most of their history, which goes back 5000 years. So the moments that I grew up in, you grew up in, where China was behind, those are actually historical aberrations. And the real norm is for China to be one of the leading world power. So they're returning to that place now. That's where they expect to be. That's where they feel like they belong. And we should definitely not cope in the same way people sometimes obviously want to cope a lot about AI and say, well, it could do this, but it could never do that. I see the same thing happening with China. And I would just say, put that idea to bed. Like they're here, they've arrived. We're not going to keep them down. We are going to have to figure out how to coexist. And ideally, work together on some things, because they absolutely have the strength where it matters to go toe to toe with us, I think, in the future, in pretty much every domain of human endeavor.