Does the human guide the horse, or the horse guide the human?

Doctorow, Cory, The Reverse Centaur’s Guide to Life After AI: How to Think About Articifical Intelligence--Before it is Too Late, Farrar, Straus and Giroux, 2026.

Doctorow’s latest book seems to want to deliver a warning about how AI is marketed to us while pointing out that there is a way to really use AI for good. The book delivers on the first part but delivers only partially on the second. This book follows closely on the heels of 2024’s Enshittification, which describes the process of a company delivering a quality product that people want and then, as they lock people in, make the product worse and worse ostensibly to make more money. AI follows a slightly different “enshittificatory” path. The whole point of AI isn’t to write, or make pictures, or make our lives easier. The point is to attract investor capital by selling those investors on a fantasy of an AI that is so powerful it can slash worker budgets by big percentages. Somehow, AI will take part of that savings from fired workers as earnings and the investors will rejoice. There’s a whole story here about what happens when masses of consumers can no longer afford shitty products, but Doctorow doesn’t go there.

He is less concerned with the dangers of AI itself than he is with its all consuming need for resources (monetary and natural). He acknowledges a use for the technology when it is a tool to help further human work. Of course, that isn’t what investors want to hear because the new tool is likely to add to costs more than it will save costs. Investors will only be willing to sink the billions of dollars they are sinking into AI companies and infrastructure as long as they think the technology will do all of these cool, money-making things down the road. The reality is, the things consumers are doing with AI, like generating clean images or writing hard emails, is primarily to keep the AI buzz going. It is the shiny ball, distracting from the fact that the companies aren’t even close to making money.

Doctorow also spends time considering the human element that AI threatens to take away from work and culture. Art, according to Doctorow, can’t be made by AI. His argument is sound: AI is just a probability engine; it is not trying to communicate anything profound--is not communicating anything at all really. Art is trying to communicate some aspect of the human condition. Further, our own legal system supports this, because Doctorow states that, for now, AI generated works lack copyright protection because humans didn’t create them. The prompt a human created can be copyrighted. If there is any art in AI, it is because a human put it there through the prompt or through engaging in an iterative process that included human decision making. Personally, I don’t care how well AI can create a story, or a song. If a human didn’t make it, I’m not interested. I allow some wiggle room. For example, I ran this essay through a local LLM and took some of its advice.

I wish Doctorow had spent more time thinking about ways we can work to shift the culture surrounding AI/LLMs. He starts to go down that road but ends by concluding that the AI bubble has to burst for real change. The challenge is to get to the bursting before our economy is so entangled with AI companies that the burst will destroy us. What he leaves unsaid but hints at is that, like climate change, it may already be too late.

x-posted and modified from a goodreads.com review.

AI disclaimer: used qwen 3.6 running in Ollama to give me some line edit and argument feedback. Some I took, some I did not.