Cory Doctorow thinks Australia should be worried about digital sovereignty

The Interview: Cory Doctorow thinks tech executives' story about AI is not an inevitability Sat 8 Aug 2026 at 5:00am Cory Doctorow is the science fiction author, activist, and technology critic who coined the term "enshittification" to describe how digital platforms seem to get worse over time as they prioritise profit. He's back with a new book, The Reverse Centaur's Guide to Life After AI, warning AI isn't a revolution rather it's a financial bubble and sooner or later it'll burst. He's visiting Australia this month, appearing at The Festival of Dangerous Ideas and The Wheeler Centre. Ahead of his visit, he tells the ABC what a life online could look like after the AI bubble and what still brings him joy on the internet. You say science fiction writers like yourself are lousy at predicting the future but they can be pretty good technology critics. Why is that? So, the point of science fiction, at least science fiction that's good — I'm sure you can find colleagues of mine who disagree with me about this, but they're wrong — is not to describe gadgets. It's to interrogate who the gadgets do things for and who the gadgets do things to. The characteristics of gadgets, obviously they matter, but who cares if there's a time machine if all it ever does is go back in time and make things worse for you. So, what's really important are the kind of techno-politics. There's a very famous example where Robert Moses, who was the famous power broker in New York City, and built a lot of its infrastructure, deliberately built overpasses on the motorway to the public beach so that public buses couldn't go under them in order to stop black people from going to the beach, right, to affect a kind of permanent segregation. And so, there's clearly some techno-politics built into that overpass, right? But it's not that the overpass is racist, right? If someone were to invent a bus with a lower profile, all of those techno politics would change. It's just a piece of architecture that's contingent on other mechanisms in society. And what science fiction writers do is they do things like imagine a lower profile bus and what might happen to the techno-politics of the things around us. And so, this makes us very well equipped to observe the claims made by tech executives, which are often very self-serving, very absolute, and very inevitable. So this story that we're being told about AI by tech companies — that it's going to change our world, It's going to replace a lot of human labour, it might even reach super intelligence. You argue in your new book that that story isn't inevitable. Why is the future still up for grabs? Well, I think we can tease apart two different things because there's two kinds of claims you made in there, one of which is just impossible, and the other one is self-serving. So, the impossible one is we teach the word-guessing program too many words and it becomes super intelligent. Yes, it's true, if you use an AI-enabled auto-complete on your phone, it can probably make good guesses about which words you're going to say. And if you're married to someone, your spouse can probably make good guesses about which words you're going to say. But the first one is just a statistical analysis, and the second one is understanding. And the fact that statistical analysis and understanding sometimes produce the same output doesn't mean that they're the same thing. And you can tell because if you turned around to your spouse and said, "I want a divorce", your spouse could have some ideas about where that thought was coming from and the chatbot wouldn't have any idea where that thought was coming from because it just wants the future to be like the past. It's doing statistical inference. So that's one half of that. The other half of that is this inevitablist doctrine that says that automation has to be driven by capital in order to improve throughput. That what you want to do is fire as many workers as possible and have the remainders pick up the slack by serving as peripherals for machines that do most of the work but then ask a human to do the parts of it that they can't do. And there's another way of arranging automation. And we see both in the historic record of how automation is adopted. When labour drives adoption rather than capital, what you see is not improvements in throughput, but improvements in quality. So, like you as a writer might choose to use a spell-checker in order to sometimes catch words that you don't know how to spell without having to stop in the middle and look them up. And your editor might say, well, now you've got a spell-checker. Why do you need a sub-editor? And so, you're going to write, we're going to publish it to the internet without an intermediary sub, and you're going to take the blame for the typos. And we can see that neither of those are built into the technology. This is about whether your boss thinks that your readers will accept a lower quality product. And we live in an environment that is the end stage of 40 years of lax competition enforcement in which cartels, monopolies, and duopolies dominate most sectors of our economy. And they believe correctly that when there are fewer options to shop for, you can lower quality without losing sales because you're the only game in town. And so, in a world of statistical inference tools and monopolies, firing workers and replacing them with software to produce lower quality outputs, is not an unreasonable bet that you could make it work, but there's nothing inevitable about it. No one came down off a mountain with two stone tablets saying "in thine future, everything shall suck". This is just the kind of end stage of a long run program to transfer value and power from labour and democratically accountable institutions to capital. It's a fear though that a lot of people have right now. But how does this outrage over AI replacing workers actually benefit big tech companies? So, we should distinguish here between AI replacing workers because bosses know that you will accept a lower quality product and so they can fire workers and replace them with AI and blame the workers for the mistakes. And AI replacing workers because they can do the workers' job. Or put more succinctly, there's two different cases. AI can do your job and an AI salesman can convince your boss to replace you with an AI that can't do your job. And I think it is absolutely reasonable to worry about the second one. But if the way that you phrase it is as the first one, AI can do my job, then you help AI companies raise money and you help AI salesmen convince bosses to fire workers. We have to really be focused on that fine distinction because it is in that distinction that the narrative that drives the investment, that drives the bubble, it's in that, that's where it lives. In your new book, you argue that what we're heading towards is not this future that we're being sold by tech companies, but instead it's the bursting of this AI bubble. So, if and when that happens what comes next? I mean it seems hard to imagine right now that AI is going away. It's already part of our lives. Hold on is it part of our lives? Because we have these digital sovereignty debates everywhere. I know it's a hot button issue in Australia. It's a big deal in Canada. I'm a Canadian, we're like cold Australians. And I was just on a panel in Toronto and we were talking about digital sovereignty and everyone on the panel, it kind of blew my mind because they all talked about how important it is for there to be Canadian AI. I was like, "wait a second, guys. If Donald Trump turned off all of our chatbots tomorrow, what would change? Which government ministry would be unable to do his job? Which important firm would go out of business?" Whereas we actually see the chief prosecutor of the international criminal court kicked off the internet. He loses his Microsoft accounts and everything else he connects with that account. If Office 365 went away tomorrow in Australia, Mad Max would come to life, right? Every ministry would cease to work, every firm would cease to function, every household would collapse. So, AI is not a part of our lives in like an important sense. AI is a part of our lives the way Tamagotchis were a part of our lives. Let's have some perspective here. Especially because there is a real digital sovereignty risk and it's nothing to do with AI. So, what would be different after the AI bubble bursts? So, we still have the machines, and you can buy them at a bankruptcy for 10 cents on the dollar, which is why if like Australia wants a national AI strategy, it should be do nothing until the hardware is 10 cents on the dollar. Jesus Christ, don't buy this shit at the top of the market. What are you, a sucker? Bubbles are really evil. They transfer money from everyday savers to insiders who lie about market opportunities. And those everyday savers are just trying to avoid like, being homeless and starving to death in their old age. And so, they do leave behind some residue. And so part of this is they trained a lot of people, some of whom were already technical and some of whom dropped out of or changed from other trades to learn all the different parts of the stack, whether it's programming, advanced statistical analysis, data labelling, prompting, all of those things have been kind of inculcated through society. And that's a little like what happened with the web. Where the web bubble took a bunch of humanities undergraduates and convinced them to become Perl programmers and HTML jocks. And so, when the web burst around 2000, all the dumb bosses who are shipping 90-pound bags of pet food saying it's the future of the web went back home. And then these workers were sitting around with servers at 10 cents on the dollar and office space going begging. And they hired each other to build real things, which we call Web 2.0 and it was pretty cool. So probably some people build it. And what will they build it with? Well, the AI companies released a ton of software for free that's open source that anyone can maintain and anyone can improve. And they did it as a kind of demo but it turns out that software is a lot more powerful than they thought it was because they never bothered to optimise it. So, in a state of nature, that software runs pretty slowly unless you throw a lot of hardware at it. But when technologists actually examine the software and see if they can tweak it to make it more efficient, they get massive efficiencies out of it. Most famously, there was a Chinese hedge fund that spun out this little skunk works called DeepSeek, and they gave them reportedly 6 million American dollars. And it was so efficient running on old chips of the sort that you can get in China, because China's embargoed for modern Nvidia chips, that Nvidia lost $600 billion off its share price in a day. You can see what's going on there. And the reason that the AI companies never bothered to optimise these old models, the reason they don't care if they're data hungry, is they're effectively running a stock swindle. And the marks, the naive investors, they don't understand the technology. Their rule of thumb is the more money they're spending, the more money they expect to get. Google wouldn't spend $100 billion on AI unless they thought $2 trillion would come out the other side. Basically, it's an investment hypothesis that goes, if the pile of shit is big enough, there's a pony underneath it. And so why bother optimising it if the way you attract suckers is by setting money on fire? So, we're going to have these models. We're going to make them run far more efficiently. We'll make incremental improvements to them. And we're not going to use it for so many frivolous and pointless things that are heavily subsidised that we need new chips every year. And it'll just become a normal technology. And if we're not careful when a third of the stock market is vaporised, when the Magnificent Seven, who are 35 per cent of the S&P 500 are all losing money on AI, when they burst. If we're not careful, we'll do austerity again. You pointed out both in Enshittification and in this recent book that the internet is really not that much fun at the moment. But is there still a place on the internet that brings you joy? Oh, I find joy all the time on the internet. It's full of weird and gnarly communities, some of which are on shitty websites. YouTube is full of great videos. I mean, the problem with YouTube isn't that it's not full of great videos. The problem is that they're siloed there. But I see wonderful things all day long. Mastodon has an algorithmic feed called Explore. I discovered Texture Tuesdays this week. And it's photographers who take pictures of really dramatic textures. I went down a rabbit hole. It was amazing. Public Domain Review just spelunks through the Welcome Trust and various state archives. New South Wales has a really good archive but also the Internet Archive and the Library of Congress highlights a couple of things every day and every day it's amazing and interesting and cool and weird. Podcasts, God, there's this podcast I love called No Gods No Mayors and the thesis is that mayor is the highest office to which a true oaf can aspire. They've done some really funny Australian mayors as well. But they do mayors through history. It's so funny. Oh my God. And one of their corollaries is that because the Pope is sometimes in charge of the bins in Vatican City, popes is mayors. So every now and again, they'll do a Pope. It's really good. You’ve said you like to start writing very early in the morning in your backyard hammock and that you have electrified undergarments and a survival so you can do it in all weather. And I had never heard of heated thermals, that sounds terrifying. Why do you like to work from a hammock? I have chronic pain and I've had a double hip replacement. It hasn't really helped and sitting upright sucks. So being able to anteriorise my pelvis in a hammock is just nice. I'm now mostly living in London. I wrote that when I was living in LA. I don't think I'm going to do all weather hammocks in London. I think it's going to be harder in the winter. But in LA in the morning, the sun's coming up and we get critters, it's raccoons and then it's opossums and then there's rats and then there's a flock of wild parrots, pigeons, many kinds of birds, hummingbirds. And just being outside, watching the sky pinking and then turn blue and it's rising over the palm trees in my neighbour's backyard and, my God, it's just so lovely. This interview has been edited for brevity and clarity. Cory Doctorow appears at Sydney's Festival of Dangerous Ideas from August 22-23 and Melbourne's The Capitol on 25 August. The Reverse Centaur's Guide to Life After AI is available now.