Cory Doctorow is not arguing that every AI tool is useless. In his view, the more important issue is how the technology is being sold, funded, and placed inside workplaces. His new book, The Reverse Centaur’s Guide to Life After AI, follows his earlier work on Enshittification: Why Everything Suddenly Got Worse and What To Do About It and turns the same critical lens toward the AI bubble.
Doctorow told Ars that he wrote the book after becoming tired of being asked to talk about AI. The result is a sharper question for readers, businesses, and investors: when AI enters a job, does it make a person more capable, or does it make the person serve the machine?
The Difference Between a Centaur and a Reverse Centaur
Doctorow uses a term from automation theory to separate two very different futures for AI. A “centaur” is a human being strengthened by a tool. That could mean machine learning, driving a car, or using autocomplete.
A reverse centaur is the opposite arrangement. Doctorow described it as “a machine head on a human body, a person who is serving as a squishy meat appendage for an uncaring machine.” In a speech last December, he gave the example of an Amazon delivery driver surrounded by AI cameras and effectively operating as a peripheral to the delivery van.
This distinction matters because the same technology can support either model. In medicine, Doctorow contrasts using AI tools to help radiologists review X-ray images with a much harsher setup: fire nine out of 10 radiologists, let AI make the diagnoses, and leave the remaining radiologist to check the machine’s work and take responsibility when errors happen.
The first version uses AI as assistance. The second version restructures the job around the machine. That is why the debate is not simply “AI good” or “AI bad.” It is about power, accountability, and who benefits when automation enters a workplace.
Why the AI Bubble Worries Doctorow
Doctorow says he uses AI tools regularly and sees real value in some of them as useful plugins or new apps. His concern is aimed at the larger story surrounding the industry: enormous capital expenditures, unrealistic expectations, self-serving messaging, and the consequences if the AI bubble pops.
“The bubble doesn’t want cheap useful things,” Doctorow said. “It wants expensive ‘disruptive’ things: big foundational models that lose billions of dollars every year.”
He argues that when AI investment mania slows, many models may disappear because running the data centers will no longer make economic sense. In his warning, “The collapse of the AI bubble is going to be ugly.”
The scale is central to his argument. Doctorow says seven AI companies currently account for more than a third of the stock market and “endlessly pass around the same $100 billion IOU.” He also says global capital expenditure was $700 billion when he wrote the book and is now $1.4 trillion.
Those numbers frame AI not just as a technology trend, but as a financial structure. The risk, in Doctorow’s view, is that useful tools get buried under a much larger speculative bet.
The Growth Story Behind the Hype
Doctorow connects the AI boom to a broader argument from Enshittification. He says firms without sufficient constraint can tilt toward worse outcomes, especially when competition falls away. Once a company saturates a market, it needs a new growth story to keep investors convinced it still has room to expand.
That pressure is especially intense for companies treated as growth stocks. Doctorow argues that growth stocks offer liquidity that firms can use to buy other companies with shares. Mature companies, by contrast, must rely on money when they want to grow.
In that context, imaginary markets become useful. Doctorow says firms can promote a story about conquering a market that does not yet have an agreed-upon valuation. If that story does not become real quickly enough, a company can move to the next one.
“The capital markets have the object permanence of a toddler, and they would lose a game of peekaboo if they were drafted to play in the league.”
Doctorow lists metaverse, crypto, Web3, and “something else” as examples of how these narratives can shift. His point is not that AI is identical to those earlier stories. He says AI is different because it is much bigger in capitalization and because there is more “there” there.
Why AI Is More Complicated Than a Simple Scam
Doctorow does not dismiss AI as empty. He calls it real computer science and points to a moment 10 years ago when computer scientists and their grad students applied existing techniques in a new way and got a surprising result. In his telling, that result initially produced returns and seemed to have somewhat linear returns on investment.
But he also says the easy gains are tapering off. There was “a lot of low-hanging fruit in AI,” he argues, and that growth period in returns to scale is ending. This creates a mismatch between the technology’s real usefulness and the scale of the investment story built around it.
That mismatch is where the AI bubble becomes dangerous. Useful AI tools may survive as practical software. The expensive, disruptive, foundational model narrative may not.
The Human Question at the Center
Doctorow also argues that AI appeals to political and business leaders because it fits a fantasy of getting things done without other people. He says people are necessary for romance, social media, board games, startups, bridges, houses, and politics. Other people also have their own priorities, which makes them difficult to control.
That is the deeper tension in his critique. AI is being marketed not only as software, but as a way to reduce dependence on human judgment, cooperation, and labor. The reverse centaur is the worker left inside that system: still responsible, still present, but increasingly organized around the demands of the machine.
For Doctorow, pushing back against AI inevitability starts with rejecting that arrangement. The practical question is not whether AI should exist. It is whether AI remains a tool people use, or becomes a system people are forced to serve.