Speaker unverified: prediction

Transcript context

…hat have you brought us today from your group chats? Okay. So, the thing that I've been putting into all my group chats because I can't make sense of it is um your guys's colleague Ezra Klein, I don't know if you noticed this, he was on some podcasts in the last month. A couple a couple. And in one of the appearances, he was being interviewed by Tyler Cowan, whose work I really admire. And then they both agreed on this fact where I was like, wait, we all agree on this fact now? where Tyler said that Sam Alman of OpenAI had um had at some point predicted that in the not too distant future we would have uh $1 billion company like a company that was valued at a billion dollars that only had one employee. Like the implication being you would train an AI to do something you just like count the money for the rest of your life. And PJ I actually believe we have a clip of this ready to go. Sam Alman says it will be possible to have billiond dollar companies run by one person. I suspect that's two or three people, but nonetheless, that seems not so far off. So, it seems to me there really ought to be significant parts of the government, by no means all, where you could have a much smaller number of people directing the AIS. It would be the same people at the top giving the orders as today more or less, and just a lot fewer staff. The reason I was injecting this into all my group chats was I was just like guys if if the conversation is among people who are quite smart and who have spent a lot of time thinking about this if they are predicting a world where AI replaces this much of the workforce this fast like how are you guys thinking about it? Every group chat I put this into, the response instead was, "What is your idea for a billion dollar company?" And any good ideas in there you want to share and maybe sort of get the creative juices flowing for our listeners? All the ideas I heard were profoundly unethical. Many of them seem to start with doing homework for children, which I don't think is a billion dollar idea, and which I think a lot of AI companies are already making money. Yeah, that company exists and it is called Open AI. It is it is a great thought experiment though. Um, you know, I I I think you know many of us have had thoughts over the years of maybe I'll go out, start a company, strike out on my own. Two of the three people in this chat actually did it. Um, but getting to a billion dollars is not trivial. And I and it is kind of tantalizing to imagine once you put AI at my fingertips, uh, will I be able to get there? Yeah. I mean, I actually this is giving me an idea for a maybe a billion-dollar oneperson startup. Uh, which is based on some of the ideas we talked about earlier in this show about how these models are becoming more uh flattering and persuasive, which is, you know, we all have that that that friend or maybe those friends who are totally addicted to posting and uh the internet and social media have wrecked their brain and turned into a shell of their former self. I know where you're going and I like it so much. And I think we should create fake social networks for these people. Oh my god, it's so good. And ir brain and turned into a shell of their former self. I know where you're going and I like it so much. And I think we should create fake social networks for these people. Oh my god, it's so good. And install them on their phones so that they could be going to what they think is X um or Facebook or Tik Tok. And instead of hearing from their real horrible internet friends, they would uh sort of have these persuasive AI chat bots who'd say like maybe tone it down with the like racism and maybe you know gradually over the course of time sort of bring them back to base reality. What do you think about this idea? I like it so much. There's so many people I would build a little mirror world for where they could just like slowly become more sane. And it's like, hey, all the retweets you want, all the likes you want, you can be like the Elon Musk of this platform. You could be like the George Dukai of this platform, whatever. But like the the the trade-off is that it has to slowly slowly make you more sane instead of the opposite. Yes. Yeah. Yes. And and I'm I worry that that is not possible because I think for you know a lot of the world's billionaires the existing social networks already serve this purpose no matter what they say they have a thousand comments saying OMG you're so true for that bestie right um and and it does seem to have driven them completely insane. So if we are able to somehow develop some anti-radicalizing technology um I do agree that could be a billion dollar company. What do you call it? Um what do you call that? I like the term uh heavenbanning which went viral a few years ago which is basically this idea that instead of being shadowbanned. Um you uh you know you would get heavenbanned which is you sort of like get banished to a platform uh where AI models just constantly agree with you and praise you. Um and this would be a way to sort of bring people back from the brink. So we can call it heaven ban. We we just spent 30 minutes talking about how when you have AIs constantly tell people what they want to think, it drives them insane. No, this is for people who are already insane. This is to try to try to rehabilitate them. Yes. I tried to have a talk with uh an AI operator this week asking it to stop complimenting me and truly it was like it's so good that you say that. Yeah. The AI always comes back and keeps trying to flatter me and I say, "Listen, buddy, you can't lick a badger twice." Okay, so move it along. Well, PJ, thank you for bringing us some gossip and content from your group chats. Uh, and we should be in a group chat together, the three of us. That sounds wonderful. Let's start one. Happy chatting, PJ. Thanks, guys. [Music]…

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