AI and the Enshittification Era with Cory Doctorow

The Weekly Show with Jon Stewart

In this episode of The Weekly Show with Jon Stewart, host Jon Stewart discusses the implications of AI with guest Cory Doctorow, author of 'The Reverse Centaur's

Key takeaways

• AI is being hyped as a disruptive force, but its real threat lies in economic disruption and potential for authoritarianism.
• The AI industry is currently operating on a model that is financially unsustainable, with high capital expenditures and low profitability.

Main topics

• AI hype and its economic implications
• The unsustainable business model of AI companies

Notable quotes

'The real threat is the economic disruption.'

Conclusion

This episode provides a critical examination of the current AI landsca

Transcript preview

Speaker 1 (0:00) Look, you know, I don't know when I'm going to leave this job. I don't know when I'm going to stay in this job. I don't know how these things are going to do. But when I do, you should replace me with Indeed. You should call Indeed when you're replacing me. Because listen, this podcast requires skills. You need a face. You need a mouth so that you can open your pie hole and just spout. That's that. Not everybody can do that. And so, you know, if you need to hire, you got to say to yourself, this is a job for Indeed sponsored jobs. You can reach candidates that meet your specific criteria, like if you want them to have skills or pie hole certifications, maybe even you want to know what the location is. Only Indeed can get that stuff done efficiently. Trust me, you know, you got to kiss a lot of frogs unless you use one of these services there like Indeed. So spend less time searching and more time actually interviewing candidates who check all your boxes. Less stress, less time, more results. When you need the right person to cut through the chaos, this is a job for Indeed sponsored jobs. And listeners of this show, we'll get a $75-sponsored job credit to help get your job the premium status. It deserves at Indeed.com slash podcast. Just go to Indeed.com slash podcast right now and support our show by saying you heard about Indeed on this podcast. Indeed.com slash podcast. Terms and conditions apply. Hey, everybody. Welcome to the weekly show podcast. My name is John Stewart, and I will be your host for today's program. It's Tuesday, July 28th. We're all recovering from the bombing raid that the President of the United States pulled on the White House Correspondents dinner. Not so fucking easy, is it? Buddy boy. It's, you know, as a comedian, you know, when you watch somebody, bombing like that, the traditional thing to say is, hey, don't quit your day job. But I think in the president's case, I think we would all make an exception. And if he would like to quit his day job and pursue full-time, I think he can improve. I think he's got some timing. The weave did quite work with some of the more concise roast. But, and once again, in the battle between Caitlin Collins and Donald Trump, Point Collins, I believe the score now somewhere around three, 128 to zero. Her sophistication focus of glass is just wiping the floor with that dude. But moving on from those events, I have been in a bit of a doomer spiral concerning AI and its effects on how it will be changing the way we work, the way we interact with each other and all those things. And I recently read a book that changed my perspective slightly. I don't want to tell you it gave me a tinge of optimism. But in fact, and it was a feeling I'm not so comfortable with, but I thought, well, geez, let me talk to this just incredibly wonderful and prolific author who wrote this book that I'm talking about. The book is actually called the Reverse Centaur's Guide to Life After AI, how to think about artificial intelligence before it's too late. So let's get to him. Let's bring on in our guest for today, Corey Doctor, Ladies and gentlemen, we are delighted and honored to have with us today. Maybe one of the most prolific writers that I know and his work is just phenomenal. You know him from Insidification. The new book is The Reverse Centaur's Guide to Life After AI, Science Fiction Writer, activist, journalist, author, for God's sakes. Cory Doctorow, thanks for joining us. Oh, what a pleasure. Thank you very much for having me on, John. The real bucket list. I'm delighted and just an enormous fan of your work in your work. writing disturbed it how much of it you're able to do in how many different genres but we're putting that aside for the moment uh you need i am an i'm a duma i believe i am a doomer and reading your book you are uh far less pessimistic not just about its ability to uh destroy our civilization as we know at the existential threat but even the problem of it as a job-destroying Cyclops that comes in and gathers up all of our productivity and puts us all on Elon Musk's dole. So let's go through. Sure. Talk me out of a tree, Corey. Yeah. So I'm really indebted to this STS Society and Technology Studies scholar at Virginia Tech. A guy called Lee Vinson coined this very useful term, which is Crap Hype. This is when you repeat the hype of the tech industry. And then at the end of it, you put in brackets, and that's bad. And the thing is, you're repeating the hype, right? You're just helping them sell it. So, you know, Mark Zuckerberg shows up one day and says, hey, everyone, I don't know if you know, but I'm like cyber-rasputin. I'm a dopamine hacking wizard where Mesmer failed and MK Ultra and pickup artists and neurolinguistic programming weirdos failed. I have succeeded. I built the mind control array out of big data. And then his critics go around and go like cyber-rasputin, the reason Grampi is a Q&ON is CyberResputin made him a Q&O. It's not because like Grampi was always racist. And Mark Zuckerberg introduced him to a lot of other racists so they could just do racist shit all day. It's because he was brainwashed by Mark Zuckerberg. And then Mark Zuckerberg's sales guys go out. And they're like, you're asking me why you should spend 40% more to advertise on Facebook? Well, I don't know. But my critics will tell you my boss is an evil dopamine-hacking wizard. and he can sell anything to anyone. Right? So this, you just don't want to do the work for them, right? Right. But is that any of that hype? So when you talk about that, is the algorithm not designed? Isn't it designed purely to do exactly what you're saying? He might not be the wizard, but in the same way that, you know, I've talked about this a little bit, in the same way that Doritos are not food. They're ultra-prud. They're designed in a lab to, kind of get past your body's ability to say, two bags, that's not enough at all. I need three. Aren't they doing that to us? So, yeah, I mean, this idea of hyper palatibility, which is like a problem with highly processed food, I see where you're coming from when we talk about social media algorithms. But I think that like, you know, this idea of Occam's Razor, you know, finding the more parsimonious explanation. Yeah, sure, like things that are entertaining or entertaining. and you'll watch them, right? But also, it's really hard to get the people you love to act in a coordinated faction. You know, economists have a term for this. They call it the collective action problem, right? You love your besties, but like the six people in your group chat can't agree on where to go for beers on Friday. So, you know, there's 200 of you on Facebook, and some of you are there because, like, you share a rare disease with a community there. And some of you're there because you, like, move from another country and that's where everyone is. Some of you're there because that's how you're organizing the carpool for your kids little league. And some of you are there because like your customers, your audience are there. And Mark Zuckerberg, you know, he kind of understands that if you love your friends more than you hate him, you'll probably stick around. And so he can turn the screws. And so you have this thing that's like, why am I on Facebook all the time when I hate it? Well, because your friends are there. Right. And you can say, well, my friends are there because, you know, they are also brainwashed. Or you can say, my friends are there because it's just really hard to coordinate our departure. Now, I've a personal example of this, which is that my grandmother was a Soviet refugee. And she had been a child soldier in the siege of Leningrad. So from like 12 to 15 was like hauling bodies. And then she gets evacuated across the winter ice, ends up in Siberia, gets knocked up by my grandfather. They desert. They go to Azerbaijan. My dad is born. And then they go all the way to Canada. You know, like all the best Americans, I'm Canadian, right? So they go all the way to Canada. No one else leaves Unangrad. They're all there because what had happened with my grandmother is she'd had like some important bond shattered, right? Like this bond where she was in this relationship where she was taking care of the older people, they were taking care of her and so on. The people who stayed behind, that's not what happened. And they all stayed. And there has not been a moment since that their lives weren't worse than my branch of the family's life. And now we get on these Zoom calls with my Russian family. And like all the men are worried about getting drafted and sent to Ukraine. And it's just awful. Right. So like, why didn't they leave? It's not because they were brainwashed by Stalin, right? They didn't leave because, like, they wanted to be with each other. And as much as they loved each other, you know, Sartre was an optimist, hell is other people. You can't get anyone to do anything ever. And so you're all just stuck there. So it's nothing more than a body at rest tends to stay at rest. And a body in motion tends to stay in motion because it does. And I certainly can't quote studies on this. But it feels like it if it doesn't rewire the mind, it is added in an accelerant into maybe some of our lesser impulses. Sure. Yeah, six and one, half dozen to the other. I mean, we have this problem all the time because, like, people exist in a society, right? So, like, yeah, we go for junk food social interactions instead of the meaningful ones. But also, like, you know, my mom's dad, who's, you know, the oldest, of 10 and dropped at a school at 13 to take care of his brothers and sisters when his dad died. By the time my mom came around and her two siblings, he was like a 35-year-old door-to-door insurance salesman who owned his own home, played bowling on the weekend, also golfed, also was like in a fraternal organization. They went to picnics and they took six weeks off on one income, right? Like, I don't know anyone who's not working like five jobs, right? So, yeah, why do we squeeze in all of our social interactions around the edges? maybe because like getting a whole evening free to go bowling and then like also being a member of like your local kwanis and whatever requires that you not be like stuck in the gig economy working every hour that God sends to pay for a house that doesn't cost 20% of your income but 60% of your income. And sure like all this stuff might be bad for you. It might be hyper palatable. You might be getting sucked into it. But like let's not give Mark Zuckerberg credit that we don't know he deserves, especially when it helps them sell ads. Right. In other words, don't, intentionality is not necessarily what he's doing. It may just be a byproduct and that if you hype them, that helps them. And that gets us to AI. Yes. Which you have kind of, the idea is AI is going into that same model. Is that the feeling that you have is we are hyping it as a planet killer, as a death star, and that is helping them in terms of valuation. Is that the mistake we're making by over blessing them? Yeah. I mean, there's lots of different stories that AI tells that are quite outlandish. So there are ways in which AI is a planet killer, right? The carbon emissions are definitely not doing the planet any favors. Actual planet killer. Right. But this idea that like, oh, my God, what if we teach the word guessing program too many words and it wakes up and turns into God and makes us into paper clips? Right? Like that is, you know, it's like, it's like, what if we breathe these horses to run so fast that one of the mares gives birth to a locomotive, right? Like, it's just not. By the way, go. Favorite sentence in the book, the reverse centaur's guide to life after AI. I have to tell you. You're talking about how they talk about AI is good. The big thing is long, long mission truckers that. Oh, long haul truckers. That. Long haul truckers. That. the AI is going to turn them into autonomous long-haul truckers, and that's going to disrupt the, you know, the largest industry in America, which is trucking, which is apparently, and you break it down very nicely that it's not. It's really like, there's only like 25,000 of them. But you say, first of all, congratulations. You've just invented a shitty version of a train. Yeah, yeah. A bunch of large vehicles following each other at a constant distance in a dedicated lane, right? I mean, you know, great. Right. That's awesome. But that's one of the greatest sort of puncturing of the bubble of its importance. And is that something we need to continue to do? Yeah. So this is another one of the stories that they tell. So there's one story that they tell, and they're all interrelated, right? But there's one story they tell the AI is just going to keep getting better and eventually it's going to be caught and it'll either solve all over problems or, you know, use us as fuel. Right. And then there's the other story, which is that the AI can do so many jobs that have such high wages that the fact that right now, like the sector is making $50 billion a year, they claim it's $60 billion a year. $10 billion of that is like money that Microsoft gave to Open AI and Open AI gave back to Microsoft. They're just passing back and forth. Yeah. Oh, beautiful. Look, I write thrillers about forensic accounting. And in my considered professional opinion, to call. call that an accounting trick is to do violence to the noble accountant trick. You know, like the CPAs of America salute you, sir. Yeah. So like $50 billion a year. When I wrote the book, the total capital expenditure to date was $700 billion. They've added at least another trillion since. And if you believe Gartner- Added a trillion in the last year. And if you believe Gartner, who admittedly did say that by now we'd all be spending six hours a day in the Metaverse, but if you believe Gartner, you know, it's going to be $2.5 trillion in total by the end of the year. So on $50 billion revenue, right? And they have to keep spending it because the data centers, the GPUs, they're like replacing them every two to five years. So like this is this expenditure just recurs. It's not like you train the model and then you're like, well, we got the model now. I guess it's all set, right? It's new model, new model, new model, right? And so you have these ongoing operational expenditures, you have these ongoing capital expenditures. And their answer to how do you turn 50 billion billion of revenue into profit when you're spending a trillion dollars a year to make it. Like, look, if you want $50 billion revenue and you've got a trillion dollars, I can do it for you. I'll give you $100 billion back. It's called just put it all in an index fund and you should be okay. I'll put it under the mattress and give you $100 billion back and we'll call it a day. Right. Right. I, you know, like Sam Malman's out there running the money losing his venture in the history of the human race. And his story about how it stops losing money is eventually you displace enough wages that bosses split the savings with Sam Altman and he comes out of hell. Right? Are you at work right now? Are you the boss? Are you the boss and you're listening to the podcast to taunt the other workers that you dominate over because you're a terrible boss? My point is simply this. This is an intervention. Stop being a control freak and doing everything yourself at work. You're busy. You're multitasking. your value. Higher experts. Upwork makes it easy to hire freelance expert specialists to get the jobs done that maybe you don't want to do or don't have the expertise to do, but don't I admit you don't have the expertise to do it, or don't have the time to do, or you want to focus on other parts of the business. Upwork is a one-stop platform. You find hire and pay expert freelancers across web and software development, data and analytics, marketing, more. Do you want to do the marketing? Do you have time to do the marketing? Do you even know how to do the marketing? What are you going to stand on a corner and pass out flyers? No, you're a busy person. That's why you're using Upwork. Then you can browse profiles, review, pass work, hire, get started quickly. I'm telling you, man, Upwork takes care of all this stuff. You don't feel like doing, and you need experts to do. And they'll handle contracts. Payments. It's free to sign up. Upwork is efficient. So visit upwork.com slash TWS right now. Post your job for free. It's upwork.com slash TWS to connect with top talent, ready to help your business grow. That's upw-O-R-K.com slash T-WS. Do it now unless you're scared. Corey, let me ask you. So, yeah, what is that when you say it generates 50 billion in revenue, right? And then you say they've added it. another 10 billion, but that's really just an accounting gimmick. What is the way that AI returns that investment? What is that 50 billion? Is it user fees? People buying subscriptions. That's all it is. Yeah, yeah, it's subscriptions. Yeah, yeah. It's enterprise licenses and subscriptions. What about government contracts and things like that? Government contracts. Yeah, all that stuff. It sums up to 50 billion. Okay. Now, there'll be new numbers and then whatever. And you may have noticed that Anthropic just said that they were profitable, but not using generally accepted accounting practices, which is just awesome. It's like, we are profitable, but the math is so cool that we can't even describe it to you. If you use regular math, it's not going to look like we're profitable. We've got this new math. And the same week, they had this other announcement, which is we have this tool, mythos, that's such a cool security tool that we can't show you how it works. It's too dangerous. But it's really good. It's too dangerous. The fact that our IPO is like, we're trying to get like this IPO in shape, right? Like, nothing to do with it. We just, like, Just whatever you, like, don't look under the blanket. What's under the blanket's amazing. It's definitely worth buying into the IPO at like 150 at, but don't look under the blanket. You look into the blanket, you're going to turn into a pillar of salt. And then, you know, it'll be on your head. So, you know, and also don't ask us to explain the math by which we're profitable. And also, we've been selling $100 bills for a dollar each. And like, when we started down this IPO where we were like, why don't we start charging $5 for these $100 bills? And every CEO who's been buying the $100 bills was like, I'm sorry. at $5, these $100 bills are not worth it. So does this then reframe is the threat of AI, not that it will turn us into paper clips or that will replace all of our high-paying jobs, but that because it is a three-card Monty, it will utterly shatter our economy. I mean, the Mag 7, most of the growth in the American economy is driven by this AI promise, right? I mean, we went through this in the Internet, boom and the internet, I think, has become standardized and is a valuable part of the economy. So is this truly a Potemkin village? What are we dealing with? So let me give you some, because that often comes up, right? This is like the web. The web lost money, AI lost money. Right. That's like, you know, that's like, you know, this food is delicious, that food is delicious, therefore they're the same food, right? Like, the fact that they share this one trait doesn't make them the same thing, right? So both go in your mouth. They both go in your mouth. Yes. Yeah. There's some Monty Python bits about this. I'm sure there is. It's a duck. It's a witch. Yeah, yeah. So here's the thing. The web had very good unit economics. What are unit economics? What is that? Yeah, yeah. So I was just going to say, no one should do an economics degree because you'll end up with an neurological injury. But learning something back at economics is useful. And so learning like what unit economics are super useful. Unit economics are what happens to your business when you get a new customer or make a new sale. Right. Just like how much money do you even? like per unit. How does your business do? Right? So in the web, like, yeah, they were losing money. But every new customer made them more profitable. Every use buy a new customer made the web as a sector more profitable. And also every generation of the web is more profitable than the previous one. Compare that to AI. Every new AI customer costs the company that just acquired the money. The more that AI customer uses their products, the more money they lose. And every generation of AI has lost more money than the previous one. Tadda-ta-ta-ta. Tease that out. So the more, because, you know, we talked a little bit about Zuckerberg and the evil genius. Clearly, the incentives on social media were an intention incentive. The more they can keep you on social media, the more money they can make, because I guess the more ads they can sell and that type of thing, is AI, the incentives are inverse to that. So the more you use their AI, how do you use their AI? that costs them more? Oh, well, because every time you use AI, it costs the money to run the inference. Can I, can I sidebar here just a little on inference? Man, anytime you want, Corey, Dr. O. Okay. Sidebar away. Okay. Because I want to mention one of the very best researchers on AI economics, which is this guy called Ed Zittron, who you should have on. Eddie Zittron. He's an amazing guy. Zed, he's British. Is Ed? All right. So Ed, someone leaked opening eyes finances to him, right? Their balance sheet, because it's all private, right? So he gets a look at it. And it really looks like their inference cost. So inference is like you type in, you know, draw me Sonic the Hedgehog with big boobs, right? And then it makes Sonic that step is the inference. Have you been reading my diary, Dr. O? Where did you? Yeah, they leak their queries. Oh, son of a bitch. Yeah. So you ask that and you got the, that's the, that's the, inference step, right? That's the using it part. In economic speak, that would be the operating expense of AI, right? So like it looks like their, their, their, their, um, OPEX is going down, which is like, okay, that's all right, because if they can just do the CAPX once, make the model and then it's, they just start making money from the OPEX, then you can make money from it, right? Just like you build a railroad and then you sweat the asset by running rail cars up and down it. But here's a thing. Their marketing budget is as big as Coca-Cola's. Now, Coca-Cola has like three full-time ad agencies, right? And ads everywhere. So where's the open AI like army of marketers? So here's Ed's theory, and I think he's right. What they're doing is they're saying, when we sell you that $100 bill for a dollar and let you do $100 worth of inference for one buy, that's not inference. That's marketing. So we're just putting it somewhere else on the balance sheet. But even, Corey, even if it's a, let's say it's a lost leader. Let's say it's, you know, they used to do that in the mall. Somebody would stand out there with a little plate with little cheese cubes and you'd walk by and go, oh, this is delicious. We'll come into Hickory Farms and they figure you'll buy, the lost leader is take this cheese and you'll buy $100 worth of Vienna sausages or whatever else is in there. But if the money that they are spending to bring in new people, but each new person who's using it costs them more money than they're making. Yeah, yeah. Well, that's the shittiest business model. And I've never even heard of anything like that. So there's two polls of finance wisdom that I turn to in these moments, right? One is this thing called Stein's Law, which is anything that can't go on forever eventually stops. The other, John Kenneth Galbraith, the market can remain irrational longer than you can remain solvent. bars. Yeah. So look, if you're spending more than you're making, you're eventually going to have a reckoning of some kind, right? But when does that reckoning come? Well, you know, like if you're Elon Musk, you take your money furnace, which is called rock, and you stuff it inside your like Beltway Bandit rocketry company that bills Uncle Sucker billions of dollars for rockets that is profitable. And then you IPO it and you get a special deal from NASDAQ so that it IPOs in the NASDAQ 100 instead of, having to wait for a couple of quarters the way it normally