Judge the Writing, Not the Tool: What the AI Pile-Ons Get Wrong

Nobody reviews the stove.

A restaurant critic sits down, eats the meal and writes about the food. Was the fish overcooked, did the sauce work, and was it worth the money? Nobody asks whether the chef used a gas range or an induction cooktop, a stand mixer or a wooden spoon. If the plate is good, it’s good. If it’s bad, the critic says so, and the brand of oven never comes into it.

Writing used to get judged the same way. Then AI showed up, and a big slice of the writing world stopped tasting the food and started inspecting the kitchen.

Why do AI pile-ons focus on the tool instead of the writing?

Because it’s easier. A book takes hours to judge, and the tool takes one accusation.

Go into almost any big writing group and mention that you used AI for anything, an outline, a blurb, a character list, and watch what happens. Nobody asks to see the book. Nobody asks whether the story works, whether the argument holds up or whether the research is right. The verdict arrives before anybody reads a page: they used AI, so the writing is crap.

That’s not a review. That’s a guess dressed up as a standard, and it’s frequently wrong.

I’ve written a whole series about what happens next, Dogpile, because the pattern repeats so often. Somebody asks a simple question. Somebody else spots a tool they don’t like. The crowd arrives, and within an hour a stranger has been called a thief, a fraud and worse, over a book none of them opened.

Why is judging the writer just as bad as judging the tool?

Sometimes the pile-on skips the tool entirely and goes straight for the person. Who is this writer, are they a real author, did they pay their dues, do they belong in the club?

That’s gatekeeping, and it’s the same mistake from another angle. Who wrote it and how don’t matter. Whether it’s any good does.

Readers have always worked this way, whether the gatekeepers like it or not. A reader picks up a book, reads a few pages and decides. They don’t check the author’s credentials, and they don’t ask what software was open on the desk. If the book holds them, they keep reading. If it doesn’t, they put it down. If people like an AI-assisted book, they’ll buy it. If they don’t, they won’t. What exactly is the crowd protecting readers from?

What should you judge a book on?

The same things we always did. Is it a good story? Do the characters feel like people? Does the article get its point across? Is it well researched, and are the facts right? Is the prose clear? Does the ending pay off what the beginning promised?

Those questions work on any book, no matter how it was made. They work on a novel typed on a manual typewriter in 1962, on a memoir written by a ghostwriter, on a business book assembled from interview transcripts, and on a book where AI did some of the grunt work. Readers care about nothing else.

Every one of them requires reading the book, and criticism that skips the reading isn’t criticism.

So the next time you watch a pile-on start, ask the crowd one question. Has anybody here read it? You’ll usually get silence, or a screenshot of a cover, or a line about how they don’t need to read it because they can tell. You can’t tell. Nobody can tell a good book from a bad one without reading it, and you can’t tell from a cover how much of the work a writer did. (See a painted cover mistaken for AI)

If you’re the writer being piled on, that question is your answer too. You don’t owe a mob a confession about your process. You owe your readers a good book.

What does misusing AI to write a book look like?

There is a way to misuse the tool.

Somebody types “write me a story about a dragon hunter,” hits enter, and publishes whatever comes back. No plotting, no characters they built, no revision, no thought. That’s having a machine produce a whole book and putting your name on it, and you’re going to get a lousy book out of it. Thin, generic, the same beats every other dragon hunter story hits, with nothing of a person anywhere in it.

The pile-on misses something about that book. It fails on its own. You don’t need to call anybody names to stop it. You read it, it’s bad, you say it’s bad, and readers walk away. The writing gets judged on the writing, and it loses. I’ve seen what happens when people pay for that kind of manuscript in She Bought a Manuscript an AI Wrote, and the problem there was the book itself.

Is a book still yours if AI did the grunt work?

Take a different writer. They spend days plotting. They build their characters, work out the backstory, decide what each person wants and what’s in the way. They use AI tools along the way, to organize notes, to check a timeline, to throw out options they’ll mostly reject. Then they go through the book chapter by chapter, and the AI handles the basic grunt writing while they shape, cut, rewrite and decide.

Whose book is that?

It’s theirs. The mind behind it is still the writer’s. Every decision that makes a story a story, what happens, who it happens to, why it matters, came from a person. AI was doing the work a virtual assistant does, or the work a ghostwriter does for a client who brings the ideas, the stories and the judgment. Nobody piles on the executive who hired a ghostwriter, and nobody should. The mind is still there.

That’s not how I write my own books, for the record. I use AI every day as a digital assistant that helps me write, and I still do the writing. But I don’t understand why another writer’s different process earns them a mob. Why every writer should treat AI as a digital assistant lays out where I draw my own lines, and other writers can draw theirs.

It’s a tool. Use it or don’t.

Is mass-produced AI publishing a new problem?

No. Book factories ran on autopilot for years before AI, and nobody raged about them.

Long before anybody had heard of ChatGPT, affiliate marketers were selling book generators. When coloring books were booming, you could buy a coloring book generator for $19.95. Then came the upsells, and the price climbed to $197. Then the courses, and you were near a thousand dollars. People bought them and turned out hundreds of coloring books a week. Puzzle books and blank journals went the same way, all pumped out by tools sold through affiliate links.

I don’t remember any vitriol about that. Maybe there was some. I never saw a writing group threaten a puzzle book publisher.

The flood is real, and the people selling the tools that cause it deserve blame. Amazon had to cap uploads because of it, and I’ve written about why KDP’s two-book weekly limit punishes the wrong authors. But the junk sinks because nobody wants it. The writer who spent a year on a novel and used AI to help organize it isn’t the flood. Treating them like it is just picks a target that can’t fight back.

Why does the anti-AI vitriol feel so out of proportion?

Read the comments under any AI post in a writing group and you’d think lives were at stake. Thief, fraud, scum. People wish harm on strangers over software, and I wrote about a thread where somebody said AI users deserve to be castrated. The temperature is that of people defending their homes.

Their lives aren’t in danger from AI. They perceive that they are.

I understand the perception, because that’s how AI gets presented to them. You’re going to lose your job. Machines will take over all the creativity. There won’t be any more art, any more writing, any more human anything. I’ve heard it all before. I’ve heard it for years. Every one of those predictions gets repeated in every thread, and the fear underneath them is real even when the predictions aren’t.

Fear doesn’t excuse cruelty, though. A frightened person who calls a stranger a thief for asking a question is still being cruel. The fear explains the anger without making the target guilty.

What does honest criticism of an AI-assisted book look like?

It looks like criticism of any other book.

Read it. Quote the passage that doesn’t work. Say the dialogue is stiff, the middle drags, the villain has no motive, the third chapter repeats the second. If the prose is full of the tired phrases and hollow transitions people associate with machine writing, say the prose is tired and hollow. Those are writing flaws, and human writers produced them by the ton long before any chatbot existed. Judge the flaw. Leave the guess about where it came from out of it.

A writer can do something with that. They can fix stiff dialogue, and they can cut a repeated chapter. Nobody can do anything with “you used AI, so it’s garbage,” because it says nothing about the book. It’s a verdict on a rumor.

Writing groups could do a lot of good here. If a group has a rule against AI posts, enforce the rule without the insults. Remove the post, explain why, and move on. If a group allows them, critique them like any other manuscript. Either way, the member who asked a question should leave with an answer instead of a pile of names.

The fall of 2026 put this argument on a bigger stage. A French novel, Thélyson Orélien’s C’était ça ou mourir, was a bestseller that critics called a masterpiece until an anonymous account on X accused it of being AI. It was pulled from a major French prize list. The same week, a detector flagged “The Blue Fighter,” the story that won Simbarashe Kundizeza the Caine Prize for African Writing, and the prize put the award under review. Both writers deny using AI.

Look at what nobody said about either book. Nobody said it was bad. Juries had read them and judged them worth a prize. Then software produced a number, and the conversation stopped being about the writing.

Will AI replace writers?

Some writing work will change, and some of it already has. It is what it is.

What it won’t replace is the person’s mind. A machine can produce sentences. It can’t decide what a story is about, why a character matters, or what a reader should feel on the last page. Even when a writer uses AI for every first draft, the mind deciding what stays and what goes is still the writer’s. That mind is what readers are paying for, whether they know it or not. The books that matter will keep coming from people with something to say.

That’s also why the tool-hunting misses. The thing worth protecting was never the absence of software. It was the presence of a person with something to say, and you find that by reading.

Focus on the damn writing

So this is my objection to all of it. The pile-ons focus on the tool. Sometimes they focus on the writer. Almost never do they focus on the writing, and readers have never cared about anything else.

If a book is bad, say it’s bad, and say why. Point to the flat characters, the sagging middle, the facts that are wrong. That’s honest criticism, and every writer can learn from it. If a book is good, it’s good, whatever helped make it. And whatever you think of the tools, drop the vitriol. Nobody’s life is in danger. Somebody’s book might be worth reading.

The rest of this series is in AI Derangement Syndrome, and the AI and Writing Hub has my longer thinking on where these tools help and where they don’t. If you’re working on a book and want somebody to judge the writing and nothing else, my book coaching service starts with a conversation.

Taste the food. Leave the stove out of it.

Frequently Asked Questions