Doctorow warns AI boom could leave workers serving machines
In a new book, Cory Doctorow argues the AI investment boom rests on weak economics and a workplace model that shifts risk onto employees.
By Maya Lindqvist · Senior Technology Correspondent
4 min read
Cory Doctorow is using his new book to challenge the idea that artificial intelligence will arrive as a force companies and workers can only accept. In an interview with Ars Technica, the tech journalist and novelist said the current AI boom combines costly infrastructure spending, weak revenue and a management fantasy of doing business with fewer people.
The book, The Reverse Centaur’s Guide to Life After AI, follows Doctorow’s earlier work on “enshittification,” his term for the decline of online platforms as companies face fewer competitive limits. Doctorow told Ars that AI has become the latest growth story for firms that need investors to believe they can still expand after dominating existing markets.
Doctorow said he does not reject AI as a field. He told Ars he uses AI tools and sees value in some applications, especially when workers choose how to use them. His criticism is aimed at the industry’s spending, labor practices and claims about replacing human work.
Centaur or reverse centaur
Doctorow frames the issue around the automation-theory idea of a “centaur”: a person helped by technology while retaining control over the task. He contrasts that with what he calls a “reverse centaur,” where a person becomes the human component of a machine-directed system.
In a December speech cited by Ars, Doctorow used an Amazon delivery driver monitored by AI cameras as an example of a worker treated as an accessory to the vehicle and its software. He also contrasted medical AI that helps radiologists spot possible tumors with a staffing model in which most radiologists are fired and the remaining doctor is left to approve AI findings and absorb blame for mistakes.
Doctorow told Ars that workers who find AI useful are often deciding for themselves how it fits into their jobs. Workers who oppose it, he said, are more likely to be asked to use it to work faster, lower quality standards or become responsible when automated systems fail.
A large bet with uncertain returns
Doctorow argued that AI spending is far larger than earlier technology manias. He told Ars that global capital expenditure tied to AI was $700 billion when he wrote the book and has since reached $1.4 trillion. He also said Meta spent $150 billion on AI over three years and has said it plans to spend another $150 billion this year, after spending $60 billion on the metaverse.
He said the sector’s economics differ from the early web. In Doctorow’s account, web services became less costly as more people used them and as the technology improved. AI companies, he told Ars, lose money on customers and usage, while newer generations of AI systems cost more.
Doctorow also warned that a sudden reversal would be economically damaging because AI companies are deeply tied to market valuations and data-center spending. In his view, many large models may disappear if investors stop funding the infrastructure needed to run them.
What might remain after a crash
Doctorow told Ars that technology bubbles can leave useful assets behind, even when they destroy wealth. He compared the dot-com crash, which left cheaper servers and a generation of web developers, with Enron, which he said produced little useful residue.
For AI, Doctorow said a collapse could leave cheaper GPUs, available applied statisticians and open-source models that others could refine. He cited DeepSeek, which Ars described earlier as spurring a $600 billion sell-off after releasing a model that ran well on commodity hardware.
Doctorow also rejected making web scraping illegal as a fix for AI’s problems. He told Ars that scraping supports socially useful archiving and analysis, and argued that labor law would do more for workers than new copyright restrictions. He pointed to Hollywood writers and actors as the rare U.S. workers who have used sector-wide bargaining to win limits on AI use.
On job losses, Doctorow drew a distinction between AI actually doing a job and managers replacing workers with systems that cannot do the work well. He told Ars there is far more evidence for the second problem so far, citing Amazon’s cashierless stores, which relied on human video reviewers rather than fully automated checkout.
This story draws on original reporting from Ars Technica.