Books take a long time. Writing, editing, and publishing one is slow, laborious work. A book that is sold today probably won’t come out until 2027 at the earliest. So it makes sense that money-grubbing innovator boys have taken aim at book publishing, which is an industry full of words, and decided that the way to disrupt it is by using AI. The dweebiest of these operations is a company called Spines, whose marketing materials feature four men wearing high V-neck branded T-shirts, looking like they are about to ask you if you’re ready for your massage. Spines is planning to disrupt the book industry by publishing—say it with me—EIGHT THOUSAND BOOKS in 2025 using artificial intelligence.
Their goal, they say, is to shrink the distance between a book being written and being published by using AI to "automate proofreading, cover design, metadata optimization and limited translation services." For $5,000, Spines claims it can have a manuscript published in three weeks. They have secured $16 million in series A funding. “We want to publish up to 8,000 books next year. The goal is to help a million authors publish their books,” Yehuda Niv, the CEO of Spines, told The Bookseller.
One of the great delusions companies like Spines are operating under is the idea that the writing of the book is an entirely different process than publishing a book. The people behind Spines keep regurgitating the same lines in interviews, about how they are not trying to “replace human creativity,” but instead leverage AI to cut through the red tape of the publishing process. “It’s a self-branding process. Let’s say you want to leave a mark, or a legacy… most people publish a book, not because they want to make a living from that. It’s their hobby,” Niv told The Bookseller.
So what they want to do is create a faster pipeline from draft to bound book. There are many steps in the process between finishing a draft and publishing a book. First, there are edits—usually, many rounds of edits, where you revise the ideas of the book and create new threads between chapters, and (generally) make the book good. Then there is fact-checking. Most major publishers do not pay for this, but it still must be done. Then there are copy edits, where all the commas are checked and spelling errors are corrected. Finally, there are several “passes” where various people have time to find problems. My new book about gossip, for example, comes out in February. By the time it is released, I will have been done with it for more than six months. We sold it in the fall of 2022.
The problem is that none of these steps are actually separate from the writing of the book. To those who buy into the myth that books are created by singular geniuses working in isolation, the publisher is the enemy. They slow down the process of letting people read your book. They gatekeep who actually gets to publish a book. But producing creative work is very rarely done by a single person. A good editor cannot be reduced to an AI-powered proofreading process. A good editor changes the way you think about the world. They show you the flaws you consistently make in singular examples so that you can improve an entire text. A good fact-checker is not just a machine; they show you flaws in your reasoning, find holes in your research, and challenge the premise of your initial ideas. A good copy editor does not just find misplaced commas; they notice all your go-to writing tics that can be broken up to improve the experience of reading. They make you better.
Even the laying-out of the book is done by human hands in publishing, because page layout does change the way a reader interacts with a book. Someone has to make sure that if a word ends up split between lines with a dash, it is still legible. A human has to make sure that all the section breaks actually wall off discrete parts of text. And even with the dozens of hands that touch every book, mistakes still slip through. Books are unwieldy, massive objects that span years of life. They are meant to take time and consideration.
That is not to say that publishing could not move faster—it should. But the delays in the publishing system are not because humans are slow and humans make mistakes. It is because the people who are in charge of these massive corporations have reduced the size of their workforce so that now their employees are doing the jobs of multiple people. The leaders of these companies decided at some point that speed was less of a priority than saving money, and now they are trying to have it both ways. HarperCollins is asking authors to opt-in to letting AI train itself on their books. In recent interview with Publisher's Weekly, the CEO of HarperCollins riffed on the vague idea of "a 'talking book,' where a book sits atop a large language model, allowing readers to converse with an AI facsimile of its author." The same article mentioned that HarperCollins is more profitable because of layoffs this year. A Dutch publisher, an affiliate of Simon & Schuster, is trying out AI for translation. Translation is a complicated and notoriously difficult art form. Languages are distinct. Some have more specific words tied to a similar meaning. They all have different idioms. You cannot just plug a book into a translator and get a Spanish-language version of a text.
“It has been a rough year for makers of original media, with the increasing use of generative AI threatening the livelihood of writers, editors, and artists of all kinds as well as the intelligence of readers,“ Maris Kreizman wrote in her top 10 books of 2024 post for New York magazine. “What a joy it is, then, to consider the books of 2024, the best of which have a specific point of view that is robotproof: a unique perspective on what it’s like to experience the world in all of its sorrow and wonder.”
Using AI to produce more books at higher volume will not help readers or writers. All it will do is flood the market with unreadable drivel that will be under-edited, poorly imagined, and riddled with errors a computer can’t catch. This is a waste of time and money, and it will make it harder to find the books that are actually good: books that are filled with heart, that have taken time and tears and difficulty to write. AI is nowhere near being able to write well or edit well. AI art sucks ass, the robots are coming for the wrong jobs, and the output is meaningless without intention. Even with all the training in the world, it’s unclear if AI will ever be able to make something that matters to anyone.
I am not actually worried that AI will ever be able to make good art. Something artificial will never be capable of having a singular experience, much less translate that into something we've never seen before. What I am worried about is how the use of AI could change the expectations on writers and authors. AI works fast because it does not have to use any critical thinking skills. It writes far faster than any human ever can, and if you are an unethical or otherwise dumb business—and there are many—you will believe that the quantity of work is more important than quality. If a publisher learns that they can get away with outsourcing more and more jobs to the machines, and if they are rewarded for that behavior by the readers, they will increase what is already a demanding pace for writers’ output.
The demand on artists to produce at the scale of an industrial factory is already reaching a fever pitch. There is more and more pressure (from fans, from the leaders of creative industries) to put out more and more work, to become a machine instead of an artist. This kind of consumption of art, and training of the consumer to always demand more, immediately, hurts everyone. It may make a few people rich very quickly, but the art will suffer. Some people can work at that speed, but not many, which means that fewer books will be made that matter to people. Fewer good stories will be told. It is stolen valor to demand the prestige of creation without the trial of actually doing it. It’s lazy and embarrassing.
This is the hard truth that no AI evangelists want to admit: You cannot hack your way into creative work. There is no shortcut, no easy route in, no infinite scale. To create for the sake of creating is an inherently human act, and to pretend otherwise is nothing more than an exercise in futility. But that doesn’t matter to the people trying to add AI to book publishing. These are the types of men who listen to history and self-help books on 3x speeds and call it brain-hacking. They don't even bother pretending they are trying to make art. It's so transparent that the goal here is profit above everything else, that even the debate itself is corrosive. Here I am, spending words on this, talking about the mechanics and business of making something rather than actually making something.
These people don’t care about creative work. They like the fact that, as Rachelle Hampton pointed out in September, “what generative AI tools like ChatGPT and Midjourney offer are shortcuts to the finish line—the rewards of being an artist without requiring any of the discipline it takes to make your art better.” The aim of AI is profit at all costs. They do not care if the product is bad, because if a bad product is the only product that exists, consumers have no choice.
But we deserve choice. We deserve writing to be protected. And it is the publishing houses that need to do this, because the AI companies never will. They need to ban AI from their businesses. They need to pledge to writers that their priority is not a check from a tech company now, but the money that can be made by holding out, by making things that are good, that people actually want to read. Great books endure, and tech trends are ephemeral. To use AI is a craven choice, but over and over again the people in power prove themselves to be just that: a huddle of cowards forever grasping outward for cash, spinning further and further from anything that ever made them relevant.
