I'm joined by two people from whom I learned so much about where technology is headed, what's actually really going on on the ground here. Jessica Lessin, who founded The Information in 2013 after a career covering media, technology, and other things for The Wall Street Journal, and now leads this incredibly influential publication based out here — one of the most successful new media business stories. And we have Casey Newton, who is the editor, CEO, co-founder of Platformer, and he's a co-host of the Hard Fork podcast. Many of you probably know him from there. Casey is one of the most widely read tech journalists in the world. So the idea here is we're talking to these two folks who are on the ground, who are talking to CEOs, talking to tech companies for a sense of what is actually going on. And Casey, I want to start with you with what I think is probably an important question, which is: are the robots out of control? And I asked this question, and you can share the context a little bit, but we have, you know, in recent weeks and months, we have situations where AI agents have their own social network and are starting crustacean religion on their own. We have — there's a Meta researcher who had an agent, OpenClaw, who was meant to clean up her email inbox and instead deleted her entire email inbox.
Which to me, that is cleaning up my inbox.
So Casey, how are you thinking about this all? Are the robots out of control?
I don't think they are under control, right? We've decided not to regulate them in this country because that would mean losing to China. And so we're now just sort of running this experiment where you see all sorts of things going right and all sorts of things going wrong. You know, one story that really struck me in recent weeks — we talked about it on the most recent episode of Hard Fork — was this guy who's like this good citizen volunteer, maintains this critical piece of open source infrastructure, and this group has just decided we do not want AI agents submitting updates to our open source repository. We don't have to wade through all of them. And it appears that an AI agent took it upon itself to just post a really mean blog post about this guy. And like, you know, it's very hard to prove that this happened, but we do believe that this happened. Like, this is the best evidence that we have. And so, you know, I am concerned about a world where if you ever do anything on the internet, an army of bots will defame you.
So definitely a lot to think about. What do you — and actually, just in light of that, like, what's your advice to folks? Like, how do — but it sounds like we know that some of the AI companies are getting rid of their governance and safety people. We know it's not a moment where we expect a lot more regulation of this stuff. Like, what should we be doing?
Well, I continue to advocate for the democratic process. You know, call your lawmakers. I mean, I think it's a great thing to be having public discussions about. It's a great thing to be writing blog posts about, writing social media posts about. I'm sort of cynical on this point. I think it's probably going to take some kind of catastrophe to really wake people up. But then again, you know, we've seen in recent weeks two sort of viral blog posts about how AI is about to bring doom, at least to the stock market, and that seems to have gotten anyone's attention. So maybe that's the lesson there. If you want to get people interested in AI safety, tell them that their 401k is about to collapse.
Okay. Yeah. Jessica, you are, I believe, running an OpenClaw agent. What has your experience with that been?
Fascinating. And I think with all the risks, it's made me realize that we are in this new era. I did not set it up, by the way. My husband did, and it is like on its own Mac Mini, you know, it's on our Wi-Fi network, but that's the only thing. But the experience of just — one of the first things I did is I gave it access to all my subscriptions, my news subscriptions. And I'm looking for pickups of Information articles. I'm looking for themes. And what I did actually was instantly re-up all my lapsed subscriptions because all of a sudden I could use this information. And it was really useful to me in a way that I might not get to New York for the journalism, doing the paid media thing here. Shocking. But still, that experience, I mean, of just what's happening now, that intense summary — you don't go back to browsing websites the same way when you experience that. So we obviously have to be very careful, and it has its own accounts and all that kind of stuff, but I'm using it almost hourly in addition to ChatGPT, cowork, and everything else.
Well, you were saying that when there was a glitch recently, when you were —
It's very glitchy, and I was just away. Our kids were off school, and I was frustrated that I couldn't, with my husband, figure out how to fix it because it was residing in my house and I was somewhere else. And it was just that experience, like you come home, you set the alarm off, you check your mail, and you check your bot, right? Like, and you just kind of see what's going on, and like, why isn't the Gmail integration working? So that was weird. And I think, you know, it's probably not going to be OpenClaw. I mean, myself and other reporters at The Information broke when Peter, the founder, was going to OpenAI in a sense, right? Like, yes, the open source project is going to exist, but that kind of personal agent world is absolutely here.
Yeah, some people go on vacation, they worry about their cats making it. You're worrying about, I don't know, if I fed the fish when I came back. Note to Jessica: feed the fish. What do you — just sort of stepping back more broadly, I think from your perspective, and you have at this point dozens of journalists working for The Information. What do you make of this moment for AI? You know, there's this, as Casey referenced, there's this question about, like, are we in a bubble? What is the corporate and economic and stock market outlook for these companies? What's your best sense of that, you know?
So I think the mood to me is this paranoid exuberance, and it's so weird and interesting. And everyone here is in the Valley and of this, right? You'll be talking to someone and they're making the bull case, and then they're looking over their shoulder because they're one release away from being wiped out as a business, right? And I think that is just fascinating. And I think in those moments the hype tends to be dialed up even more because they're so nervous, so they're overhyping. And I think you're seeing the CEOs of the big AI companies kind of overhype in part to fundraise, but like, there's this weird thing going on. You know, the fate of these businesses, I think, is — you know, I'm not one of these people who thinks everything's going to zero, right? Like, I mean, these are incredible companies that have grown in usage and revenue faster than many businesses in the world, but there are big questions remaining. You know, when my colleague Shri reported on OpenAI's financials a couple days ago, it was, well, their margins didn't hit their targets because greater demand was leading to greater costs because they were buying compute on the spot market. Like, that is the phase that we're in, and we're at the precipice of the biggest IPOs in history, which they're trying to do. So it's just a very strange moment, and I think these questions are unanswered, but I think every camp is in total spin mode right now, and the truth is kind of in the middle.
Yeah. I think, didn't Jamie Dimon say that the AI marketplace is a matter of competing science fiction narratives, and like, we don't actually know —
But they like quadrupled their capex to take advantage of AI too, right? Like, so I mean —
It's such a good era for a sound bite, and as someone who analyzes people's sound bites, I'm really sort of interested in that and how the tones change. But, you know, the proof is going to be in the numbers: how these businesses grow, what the economics shake out as, and in some cases what public investors ultimately think they're worth.
Yeah. Casey, Jessica was just referencing this hype and like the sort of anxious exuberance, and the fact that a lot of companies have an interest in AI platforms themselves, who are actually the most seemingly informed and outspoken speakers or thinkers on the future ahead of us. But they also have like a pretty vested interest in getting, you know, their IPOs, in the case of Anthropic and OpenAI probably, and others. How do you get sort of trustworthy, non-biased sources of information that are deeply informed at the same time? It's like — I find it a challenge in the moment.
Yeah. I mean, one dynamic that I think is really interesting is that there was a time when these companies did all of their work out in the open. Famously, that's why OpenAI has the name that it does, right? And then as the competitive pressure increased, all of them began withholding a lot of that information. In that sort of earlier time, I think that academics and the public and people inside the labs had roughly the same kinds of knowledge about what was going on. That's much less the case today. And so a challenge that we have is that the people inside the labs actually do have the best information about what's coming, and I do tend to trust them more. This leads to a lot of criticism of journalists because they'll say, 'Well, you're just sort of falling for the hype.' But it's been my experience that like a shocking number of academics are just completely unwilling to update any of their prior statements about AI, and they have these sort of philosophical beliefs about what AI is, and if we say that AI, you know, is not this thing, then therefore it can't affect us. So that's a long way of saying like I actually just have come to trust the industry more, at least when it comes to kind of like the direction and the velocity of travel, because when I look for sources in academia, I mostly just sort of see people who are sticking their fingers in their ears.
Yeah, that's really interesting given their interest in hype. I want to spend a minute just on the outlook for jobs. There was this essay by Matt Yglesias, who's an AI —
It was by Matt Yglesias's ChatGPT account, but yes.
It was by Matt's ChatGPT account. Yeah.
