Doctorow's Reverse Centaur, Reviewed From Inside the Agency That Sells AI

Doctorow's new book is for the people who buy AI. I'm the guy those people hire to use it.

He's mostly right about the money. Mostly right about the labor. He's got almost nothing on the rooms where the actual decisions get made.

The Reverse Centaur's Guide to Life After AI landed in June 2026. Goodreads has it at 3,805 ratings and 572 written reviews, averaging 4.09.

That average buries the part worth reading. Under it sit 105 one-stars and 145 two-stars, with another 500 readers parked at three.

About a fifth of the people who bothered to rate it came away unhappy, and not for the same reason. Some say he's too soft on AI. Some say he's too hard. A few just think he's loud.

That spread is the useful thing about the book. It's why the reviews matter as much as the text.

The frame, and where it actually holds

A centaur runs the machine on their own terms. A reverse centaur gets conscripted to keep pace with a machine already running the show. Doctorow named that clean, and I watch it happen every month.

His line, the one nearly every reviewer copies down, is "the most important aspect of a new technology isn't what the machine does, it's who it does it for and who it does it to." A four-star reader put that next to the sharper one: "Never forget that you aren't the target for AI hype (investors are)."

Not everybody buys the metaphor. One of the thorough four-star reviews said it flat: "I'm not sure this metaphor is as good as the author thinks, but we are going to have to run with it, given it's the title of the book." Fair. The horse imagery gets thin by the middle of the book. The underlying distinction still earns its keep. It explains something I couldn't name before.

I shoot and write for food, beverage, and CPG brands. Small ones, mostly. A tortilla company, a tire company, a bottled water brand, a boating channel. Nobody on that list has an AI strategy. Every one of them has a budget, a shelf date, and a marketing manager who got told upstairs to do more with less.

When I use a language model to knock a 900-word brief into shape before I go shoot the thing myself, I'm the centaur. The machine does the part I hate. I'm still the one deciding what the picture means. When a client's marketing manager gets handed a quota of forty pieces a week and a seat license, then spends her Thursday cleaning up slop she didn't ask for, she's the reverse centaur. Same software. Completely different life. Doctorow is right that the difference isn't technical. It's who signed the purchase order.

The blind spot a three-star reader found before I did

The sharpest criticism on the whole page came from a reader who gave the book three stars and clearly liked it. "The author understands well the incentives of technology companies, their investors, Labor, individual AI consumers, and American finance. His only (yet critical) blind spot is the business consumer and board rooms of the small and medium sized American businesses that use and make bets on AI. He's very focused on the top of the economy at the expense of missing a story on how a huge group of people are experiencing AI."

That's the entire gap. It's my whole client list.

Doctorow's villains are hyperscalers, venture funds, and the investment bankers pricing a $16 trillion fantasy. Fine. Those people exist and they're doing what he says.

The AI decision that wrecks a marketing department isn't made in Menlo Park. It's made by a VP at a regional food brand who read a LinkedIn post, watched a competitor post twelve reels in a week, and came back from a trade show convinced he was behind.

Nobody in that chain has heard the phrase "Byzantine premium." They're not investors. They're scared.

Doctorow's book has a great theory of the con and no theory of the mark.

He also coins terms that deserve to escape the book. "Automation blindness," the way workers under automated systems lose the skill to judge the output they're supposed to be checking. "Accountability sink," where the human in the loop gets blamed for the machine's failure instead of the executive who bought the machine. "Vocational awe," which he defines as the process by which employers exploit their workers' sense of duty to the people they serve to get them to accept brutal working conditions.

I've seen all three in food marketing. Vocational awe is the load-bearing wall of every underpaid content job I've ever watched somebody burn out of.

Where the angry readers have a point

The one-star pile isn't all bots and grudges.

One reviewer landed the best line on the page: Doctorow "makes some valid points," but "his delivery has the energy of a Karen who's eggs are runny at Dennys and wants to speak to the manager." That's mean. It's also not wrong.

Another two-star reader, clearly no fan of the politics, gave him the substance and then drew the knife: "this guy hates tech companies, corporations, basic economics, finance, and venture capital more than any author I've ever read. It's like 75% left wing talking points, but there are a few nuggets buried in here if you can sift through all the propaganda."

The serious objection comes from the other direction. A two-star who seems to work next to the field wrote that Doctorow "is proudly incurious about the capabilities of LLMs," that his habit of calling them next word predictors "leads to an embarrassing underprediction of their development," and that a year after it was written, "it already feels incredibly dated." He closed with the line that should worry the publisher: in this book "we are stuck with a stubborn Gen Xer fighting the previous war."

I don't fully agree. I get the complaint. I use these tools every working day on real deliverables. The 2025 version and the 2026 version are not the same animal, and a book that shrugs at the difference is going to age like milk in a Marfa parking lot.

Doctorow is arguing the business model, which moves slow, off the capabilities, which move fast. That's a structural risk he took on purpose. The reviews are already billing him for it.

Then the craft complaint shows up over and over. One three-star: "this book felt disjointed to me, like he was stitching together blog posts." Another: "It doesn't really build up its ideas properly, but instead interrupts them with long asides before even finishing the original thought." And the one that made me laugh out loud, from a reader mad about the missing citations: "It's ironic that a book that spent so long arguing for fair recognition and compensation for writers took so little effort to reasonably attribute cited work."

So who should actually read it

Five-star readers aren't wrong. One of them described the audiobook this way: "It feels like we got together for a coffee, and I asked him to help me understand what is actually going on with the current AI bubble." That's the real product. It's 240 pages of a very well-informed friend explaining the financing, and it does that job better than anything else on the shelf right now.

If you run marketing for a food or beverage brand, read the labor chapters and skip the rest. Not because the rest is wrong. Those labor chapters are the part you can act on this quarter.

Ask who at your company chose the AI tool. Ask whether the person using it every day had a vote. Ask what happens to that person when the output is bad. If she gets blamed, you've got an accountability sink. You didn't need a trillion-dollar bubble to build one.

And if you're the person buying the seats, Doctorow didn't write this book about you. He wrote it about the people selling to the people selling to you. Somebody still needs to write the one about the trade show floor in Dallas where the actual damage gets ordered.

The book is worth your five hours. Just don't mistake a very good map of the casino for a map of your own store.

Sources: The Reverse Centaur's Guide to Life After AI by Cory Doctorow, MCD, June 23 2026, 240 pages. Ratings, review counts and all reader quotes from Goodreads as of September 29 2026. Reader quotes are short excerpts used for commentary and belong to their authors.