Leopold Aschenbrenner: preference

Transcript context

…That will be interesting. Going back to your strategy about how important the 2030s will be for how the rest of the future goes, you want to be in a position of influence by that point because of capital. As far as I know, there's probably a whole bunch of literature on this, I'm just riffing. The landed gentry before the beginning of the Industrial Revolution, I'm not sure if they were able to leverage their position in a sort of Georgist or Piketty-type sense, in order to accrue the returns that were realized through the Industrial Revolution. I don't know what happened. At some point, they just weren't the landed gentry. I'd be concerned that even if you make great investment calls, you'll be like the guy who owned a lot of farmland before the Industrial Revolution. The guy who's actually going to make a bunch of money is the one with the steam engine. Even he doesn't make that much money because most of the benefits are widely diffused and so forth. The analog is you sell your land and you put it all in the people who are building the new industries. The real depreciating asset for me is human capital. I was valedictorian of Columbia. The thing that made you special is you're smart. In four years, it might not matter because it's automatable. A friend joked that the investment firm is perfectly hedged for me. Either AGI happens this decade and my human capital depreciates, but I turn it into financial capital, or no AGI happens and the firm doesn’t do well, but I’m still in my twenties and smart. Dwarkesh Patel 04:17:44 Excellent. What’s your story for why AGI hasn’t been priced in? Financial markets are supposed to be very efficient, so it’s hard to get an edge. Naively, you might say, “I’ve looked at these scaling curves, and they imply we’ll be buying much more compute and energy than analysts realize.” Shouldn’t those analysts be broke by now? What’s going on? Leopold Aschenbrenner 04:18:10 I used to believe in the EMH guy as an economist. But now, I think there are groups of smart people, like those in San Francisco, who have alpha over the rest of society in seeing the future. It’s like with COVID. A similar group of people saw it coming and called it completely corrected. They shorted the market and did really well. Why isn’t AGI priced in? It’s like asking why the government hasn’t nationalized the labs yet. Society hasn’t priced it in yet. It hasn’t completely diffused. I might be wrong but not many people take these ideas seriously. Dwarkesh Patel 04:19:14 There are a couple of other ideas I was playing around with that we haven’t gotten to talk about yet. One’s systems competition. One of my favorite books about World War II is Victor Davis Hanson’s summary of everything. He explains why the Allies made better decisions than the Axis. Leopold Aschenbrenner 04:19:39 Why did they? There were decisions the Axis made that were pretty good, like blitzkrieg. Leopold Aschenbrenner 04:19:45 That was sort of by accident though. Dwarkesh Patel 04:19:47 In what sense? That they just had the infrastructure left over? Leopold Aschenbrenner 04:19:49 My read of it is that blitzkrieg wasn’t an ingenious strategy. Their hand was forced. This is the very Adam Tooze-ian story of World War II. There’s the concept of a long war versus a short war, which is important. Germany realized that if they were in a long war, including the United States, they would not be able to compete industrially. Their only path to victory was to make it a short war. That worked much more spectacularly than they thought, allowing them to take over France and much of Europe. The decision to invade the Soviet Union was related to the western front because they needed resources like oil. Auschwitz was actually a giant chemical plant to produce synthetic oil and other materials. It was the largest industrial project in Nazi Germany. They thought, “we crushed them in World War I, it’ll be easy. We’ll invade, get the resources, and then fight on the western front.” Even during the invasion of the Soviet Union, even though a large number of the deaths happened there, a large fraction of German industrial production—planes, naval forces, and so on—was directed towards the western front and the western allies. By the way, this concept of a long war versus a short war is interesting, especially when thinking about the China competition. I worry about the decline of latent American industrial capacity. China builds like 200 times more ships than we do right now. Maybe we have superiority in the non-AI world in military materiel and can win a short war or defend Taiwan. If it drags on, China might be better able to mobilize industrial resources in a way we can’t anymore. This is also relevant to AI. If building AGI requires a trillion-dollar cluster instead of a $100 billion cluster, or even if it’s on the $100 billion cluster, it really matters if you can do an order of magnitude more compute for your superintelligence. Maybe right now they’re behind, but they have the raw latent industrial capacity to outbuild us. That matters both in the run-up to AGI and afterward. You have the superintelligence on your cluster, and then it’s time to expand the explosive growth. Will we let the robo-factories run wild? Maybe not, but maybe China will. How many drones will we produce? There’s an industrial explosion that I worry about. Dwarkesh Patel 04:22:41 You’ve got to be one of the few people in the world who is both concerned about alignment but also wants to ensure we let the robo-factories proceed once we get ASI to beat out China. That’s very interesting. Leopold Aschenbrenner 04:22:54 It’s all part of the picture. Dwarkesh Patel 04:22:59 Speaking of ASIs and the robot factories and robo armies.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.