In a wide-ranging interview, tech journalist and science fiction author Cory Doctorow dissects the AI bubble, arguing it is the most expensive and dangerous speculative mania in history. His new book, The Reverse Centaur's Guide to Life After AI, serves as a follow-up to his earlier work Enshittification, offering a framework for understanding why the AI industry is built on unsustainable hype and what might be left after the crash.

The Roots of the Bubble

Doctorow explains that the AI bubble emerges from a structural need for growth stocks. When tech giants saturate their markets, they must invent imaginary markets to sustain the illusion of growth. Unlike previous bubbles—such as the dot-com boom—AI requires astronomical capital expenditure, currently nearing $1.4 trillion globally, while generating only about $50 billion in annual revenue. "AI is the money-losingest thing our species has ever done," Doctorow says, noting that every customer and every use case deepens the losses.

The ideological appeal of AI lies in a "fantasy of a world without people." Powerful leaders, frustrated by reliance on human cooperation, embrace AI as a tool to eliminate workers and dependence on others. This fantasy, combined with the material need for a growth narrative, has fueled the largest misallocation of capital in history.

Centaurs and Reverse Centaurs

Doctorow introduces the concepts of "centaurs" and "reverse centaurs." A centaur is a human augmented by technology on their own terms—such as a radiologist using AI to spot tumors. A reverse centaur is a human reduced to a "squishy meat appendage" for a machine, like an Amazon delivery driver monitored by AI cameras. The AI industry predominantly aims to create reverse centaurs, forcing workers to serve algorithms while taking blame for errors. This explains why many workers resist AI while others embrace it: the difference lies in who controls the technology.

The Economic Catastrophe Ahead

The scale of the AI bubble dwarfs previous manias. Seven AI companies now account for over a third of the stock market, passing around the same $100 billion IOU. Doctorow warns that when the bubble bursts, most models will disappear because data centers will become uneconomical to operate. However, he draws a distinction between bubbles with "productive residues" and those without. The dot-com crash left behind cheap servers, skilled programmers, and open-source technologies that enabled Web 2.0. The crypto crash left only "shitty monkey JPEGs and worse Austrian economics."

Doctorow expects the AI crash to produce useful leftovers: cheap GPUs, a glut of applied statisticians, and open-source models that can run on commodity hardware. He points to DeepSeek, a Chinese hedge fund spin-off that optimized open-source models with just $6 million, triggering a $600 billion market sell-off in 24 hours. "If you've got cheap hardware and you've got applied statisticians, you've got these open source models… that's a better setup than one in which we're all running around arguing about whether the word-guessing program is going to become God."

Why Workers Wanted Tech Before but Reject AI

Unlike earlier technological breakthroughs—such as the web—which workers smuggled into workplaces because they saw clear benefits, AI is being imposed on employees. Bosses use surveillance software to punish workers who refuse to adopt AI. This top-down imposition, combined with the technology's poor unit economics (each use costs the company money), creates resentment. Doctorow notes that workers who find AI helpful are those who can decide how to use it; those forced to use it as an "accountability sink" are miserable.

The Fantasy of a World Without People

Doctorow emphasizes that the drive for AI reflects a deep-seated desire among the powerful to eliminate the friction of dealing with other humans. "Hell really is other people," he says. Corporate leaders fear that they are not actually driving the company—the workers are. AI offers the illusion of control: they can "wire the toy steering wheel directly into the drivetrain" and bypass ego-shattering confrontations with knowledgeable employees. This ideological attractiveness, combined with financial necessity, explains the extraordinary scale of investment.

Scraping and Copyright: A Dangerous Battle

Doctorow pushes back against calls to ban web scraping for AI training. He argues scraping is socially beneficial—it enabled the Internet Archive to preserve lost news content, for example. Prohibiting scraping would harm everyone except big media companies, which can license their corpuses to Big Tech. Instead of new copyright laws, Doctorow advocates for new labor laws, specifically sectoral bargaining. He points out that Hollywood screenwriters and actors, who are exempt from Taft-Hartley restrictions on sectoral bargaining, successfully beat back AI threats. Extending such bargaining rights to all workers would protect them from being replaced by algorithms.

When the Bubble Bursts: What Remains

Doctorow advises investors to "go long on laser tag arenas" because data centers can be repurposed for little else. Yet the residue of the AI bubble could be valuable: open-source models, trained statisticians, and cheap GPUs could enable genuinely useful applications. He cites his own use of local AI models for transcription and typo detection as examples of benign, low-energy tools. His friend Patrick Ball at HRDAG uses Copilots to help exonerate wrongfully convicted prisoners by prioritizing cases with linguistic correlates to past exonerations. These applications are "good, and the proof is in the pudding."

Doctorow distinguishes between AI actually replacing jobs and bosses being foolish enough to believe it can. He offers a wager to AI executives: when they retire, would they want their elderly care provided by a person or an AI? The answer, he suggests, reveals the hollowness of the hype. The real challenge is not technological inevitability but political choice—whether we build a society where AI serves humans or one where humans serve machines. The collapse of the bubble will be painful, but it also offers an opportunity to salvage the useful bits and discard the rest.

Source: Ars Technica News