Essay: The Impact of AI on the Book Publishing Industry

In this book I’ve generally considered AI’s short-term impact on book publishing. That’s evolution. But is a revolution in the offing?

You’ll recall Doc Searls’ quotation from the introduction to this book: “The next revolution will catch us all off guard, as they always do. Said another way: if the crowd is anticipating the revolution, it can’t be the revolution.”

By that statement I would likely fail if my prediction was for an incipient AI revolution. Nonetheless, I want to play with the possibility over the next 20 pages or so.

What got me started was a 2023 conversation with Peter Brantley, my partner in the AI webinar weeds, about how we might shape upcoming AI programs. Peter works much of the time within the library community, and had just returned from a in-person day-long event in Los Angeles with some 150 of his library colleagues, where they explored AI’s potential impact on their sector. The conversation was lively and in-depth. Why, he wondered, don’t trade book publishers have gatherings like this? AI is a topic at every publishing event: why isn’t it the topic?

The answer Peter came up with is that trade book publishers haven’t yet concluded that AI is going to have any real-world consequences on their business. On the core of their business. On their ability to conduct business. Sure, it might help write a press release or pen a manuscript rejection letter, but they assume that publishing will still putter along as it’s been doing for decades, with the average time from contract to print book dragging on for up to two years.

At the same time, publishers of course face near-term challenges more pressing than AI. Paper prices. Shipping costs. Shrinking margins. Sales mostly flat overall. With all of those pressures, who has the time or bandwidth to care about AI?

But the other creative industries get it. Art and design gets it. Advertising gets it. Hollywood sees what’s coming, as does the music industry. Journalists are watching with dread.

Other book publishing sectors are starting to get a handle on AI’s impact. Scholarly publishing is deep into the technology. But trade publishers are acting like there will be few real-world consequences from AI.

What might these consequences be? Will they impact the supply of books? Or the demand for books?

Supply: Yes, AI is the culprit behind a bunch of new garbage books on Amazon. Hundreds? Certainly. Thousands? Probably. Millions? No. How many books are available on Amazon in all formats? More than 50 million. It’s very crowded already.

That aside, it’s clear that in the near term AI isn’t going to be writing whole books that people will actually want to read. Its value thus far is as a writing buddy—critiquing, suggesting, occasionally spinning out a few paragraphs of usable text. No, in the near term, AI is not going to be radically changing the outputs of book publishing, the supply of new books.

Demand: No one is demanding AI-generated books. That’s not a factor. And I can’t foresee any scenario where AI will impact the demand for books more broadly.

What about the process of publishing? Yes, that will change. Publishers are now looking to AI for help here and there: marketing, editorial, a bit of this and a bit of that.

Is it possible that there will be few real-world consequences of AI on trade book publishing? Perhaps it’s just a toy.

So let’s step back from the AI weeds and revisit the troubled state of trade publishing, and through that, drill down into where AI might have its greatest impact.

By varying estimates, the entire book publishing industry in the United States has annual sales of under $35 billion. Apple Computer’s annual sales are more than 10 times higher. Apple’s gross profit is some 47% of sales, alongside net profits of 24% of sales. Publishing doesn’t even dare dream of margins like these.

Trade book publishing has been in a gentle economic decline for decades, some years up a few percent, some years down—overall, trade book publishing is not a growth industry.

In 2024, according to AAP’s StatShot “trade revenues were up nearly 5%, at $8.578 billion for the calendar year.” An up year, after a decline of 0.3% in 2023.

Without the growth of alternative formats, ebooks and audiobooks, the industry might be in very bad shape. Ebooks were a growth engine for years. Audiobook sales continue to climb. The latest data (as of May 2025) shows sales up nearly 15% in 2022 and another 9% in 2023. Yes, some of that is substitution sales, but a lot of the customers for ebooks and audiobooks are new customers, not necessarily regular book readers. In 2024, ebooks and audiobooks accounted for over a quarter of trade sales. Would all of those customers buy print books if print was all that was available?

Retail pricing is becoming an increasing concern as costs and resale discounts squeeze publisher margins. Academic studies suggest that there is a degree of price elasticity for books, but surely we’re reaching a resistance point: hardcover bestsellers are running up against what may be a $35 price ceiling. “Subscription fatigue” is causing video-on-demand customers to drop services, focused on the increasing cost of each. Netflix’s Premium plan now runs to $24.99/month; the Disney Plus Duo Premium is $19.99/month. As a colleague recently remarked to me, how many cash-strapped subscribers will say, oh, let’s cancel our Netflix subscription so that we can buy a book next month?

I’m not going to beat this to death: it’s not only authors who scramble to make a living—publishing personnel are grossly underpaid compared to other white-collar professions.

I keep a representative smorgasbord of recent publishing job postings. I’m looking at one, for a Marketing Manager at a trade publisher that reported 2024 sales of just over $30 million. The position pays between $60,000 and $70,000 per annum. Here’s another, for a Publicity Manager at one of the imprints of a publisher reporting $750 million in annual sales. “This role will have an annual salary of $74,000–$79,000,” the listing notes.

According to the U.S. Bureau of Labor Statistics National Occupational Employment and Wage Estimates, marketing managers earn, in the mean, across all industries, $166,410. The mean earnings for “Public Relations Managers” are $159,420.

Book publishing isn’t even in the ballpark. This is not news to anyone working in publishing. It’s always been this way. But can publishing sustain itself at these pay levels? An inability to offer candidates even half of the pay standard does not auger well for an industry growing increasingly digital.

In an age of TikTok, YouTube, social media, SEO, metadata, and author platforms, what the heck is a “publicity manager” anyway?

Three persistent advantages for traditional book publishers are (i) cachet, (ii) access to major media, and (iii) access to bookstore distribution. This doesn’t apply equally to all traditional publishers—the largest ones have more of each: more cachet, better access to major media, and better opportunities for getting their books onto the front tables at the largest number of bookstores. But, taken together, these are the characteristic advantages of the traditional book publishing industry.

One other value that publishers sometimes offer is a curated list of complementary titles, where each single title benefits from its companions. Think of a publisher like Fox Chapel, specializing in books about crafts: if you enjoy one of their titles, there’s a reasonable chance you’ll look closely at their other books (and magazines) on the same topic, even if they’re written by other authors.

This crosses over to something that only a few publishers have managed to develop, a marketable brand. Think of series like Wiley’s “For Dummies,” or Oxford’s “Very Short Introductions.” The brand conveys a consistency of editorial approach and quality that encourages readers to buy multiple titles. Self-publishers often launch their own short series, mostly for fiction, though they can’t inhabit the breadth of these larger publisher catalogs.

All of the other services that traditional publishers provide can be purchased on the open market, with similar quality, at affordable prices (for example, editorial, design, production and essential marketing functions). And, for self-publishers, the per-book-sold income is five times or more greater than traditional royalty schedules.

Let me quickly demean the value of the top three advantages. Cachet is good for bragging rights, but has little cash value. Book reviews, and other major media exposure, which have always been tough to get, now have a vastly diminished impact on book sales. And bookstores represent roughly 15% of industry sales (and far less for self-published authors)—they are no longer core to a book’s success.

Things have changed.

Self-publishing has been the main driver of growth in consumer (trade) publishing for the last decade and more. Because Amazon is demure, accurate sales data around self-publishing remains hidden—it’s easy to ignore its importance. But the available data shows that self-publishing claims a substantial portion of the trade publishing market.

As Kobo CEO Michael Tamblyn (somewhat) famously noted, “One in 4 books we sell in English is a self-published title, which means that effectively, for us, self-publishing is like having a whole other Penguin Random House sitting out in the market that no one sees. It’s like the dark matter of publishing.” (Penguin Random House’s 2024 worldwide sales were just over $5 billion; U.S. sales were $3.2 billion.)

The trends within the self-publishing market are arguably the only significant trends in trade publishing—self-published authors are showing the way. They are adventurous and uninhibited by the customs of the traditional industry. As an example, traditional publishers still shy away from using AI for audiobooks. Self-published authors show no such reservations.

Many are fluent with social media. They are close to their readers, their customers. If you want to learn the latest marketing techniques, follow author forums, blogs, and newsletters.

Dig deep into author income stats, the economics of writing, and then compare the income decline of traditionally-published authors against the income gains of self-published authors. In an international survey and report that I conducted with Steve Sieck for ALLi, in 2023 (pdf), we discovered that “the median writing and self-publishing-related income in 2022 of all self-publishers responding was $12,749, a 53% increase (italics mine) over the previous year. Average (mean) incomes skewed much higher: $82,600 in 2022, a 34% increase.”

Does that sound high? In a separate survey, conducted by Peter Hildick-Smith for the Authors Guild in 2023, “full-time self-published authors, who had been publishing since at least 2018, reported a mean income of $24,000 compared to $13,700 in 2018, a 76 percent increase (italics mine).”

In part for the sake of thoroughness, I want also to mention hybrid publishers, who combine the freedom of self-publishing with some of the rigor of traditional publishing.

It can be expensive, but I’ve become increasingly enthusiastic about the hybrid model, and the hybrid publishing segment is significant and growing. No one estimates hybrid sales separately. But the impact is most clearly seen in its dominance of two categories: book by celebrities, and in popular business books. These categories have long been reliable income sources for trade publishers, but hybrid publishers’ “concierge services,” and much more favorable income sharing, are irresistible for many high-profile authors.

Recent developments, like Authors Equity’s profit sharing relationship with its authors, and Keila Shaheen’s 50/50 profit sharing with Simon & Schuster, suggest a trend toward the normalization of the hybrid model. This is good news for authors; less promising for traditional publishers.

If AI is to transform book publishing, it’s going to have to move beyond the book.

I’m co-authoring a report with Rüdiger Wischenbart, based on his original study “Publishing Beyond Publishers.” We’re trying to understand, and to quantify, as best we can, all of the book-ish publishing activity, worldwide, that could (and should) be included as part of a more comprehensive view of the modern publishing ecosphere.

Once you move beyond the bookish container you find glimpses of where AI could really make an impact.

The poster child for our project is Wattpad, where “97 million people spend over 23 billion minutes a month engaged in original stories.” The stories are experienced online, mostly on smartphones, in short chunks. Few become published books. They are just as likely to find their way into “Wattpad WEBTOON Studios, the company’s TV, film, and publishing counterparts.”

Younger readers are more attuned to online digital reading than their more senior brethren. According to a January 2024 Wattpad survey, “digital formats are increasingly popular among younger generations, with 65% of Gen Z and 71% of Millennials embracing webnovels, ebooks and webcomics, while less than half of the Gen X and Boomer generations say the same.”

And, of course, as the traditional publishing industry is now learning: “Diversity is a key driver for embracing digital formats: 61% of Gen Z and 70% of Millennials agree that ebooks, webnovels, and webcomics give them access to content that’s harder to find in bookstores and libraries, including LGBTQ+ and minority-focused content.”

Wattpad has competitors. Several are nonprofit and open source. Inkitt is the top commercial competitor, which, according to Publishers Weekly, recently raised $37 million from investors “including Stefan von Holtzbrinck, the owner of Holtzbrinck Publishing Group, which owns Macmillan, and former Penguin CEO Michael Lynton; it has now pulled in a total of $117 million in investment. In February 2023, the Financial Times said that Inkitt was the eighth fastest-growing company in Europe, and #1 in Germany.” A July 9 article in Esquire outlined Inkitt’s use of AI, its role in their success.

How many of you have even heard of Inkitt, visited its publishing platform, or downloaded its Galatea reading app?

But our “Publishing Beyond Publishers” report goes beyond online story platforms to consider how “content can be created and disseminated in a diversity of formats (print, digital), media (books, audio, movies, games), distribution channels (communities, platforms, streams) and business models (sales of products, subscriptions, streaming, freemium, paid models) in mostly digitally-defined supply and marketing chains.” How many opportunities are book publishers missing because they’re not an obvious fit within the current business model?

With the exception of ebooks, modern publishing has never faced a serious threat from technology. Bookish apps appeared to be potentially disruptive, but only briefly. The dawn of the Internet era offered publishers more opportunity than threat; it’s just a single retailer, Amazon, enabled by the Internet, that truly upset the apple cart.

I’ve never seen a study on whether Amazon has been (when all aspects of its operations and impact are fully factored) a net positive for the book publishing industry. It reaches many buyers who aren’t served by bricks and mortar, and is willing to sacrifice margin to keep prices low. Ebooks and audiobooks are delivered at scale. But Amazon is able to cut prices in part because it demands high discounts and fees from suppliers, and it chokes other retail channels. There are painful trade-offs.

Still, don’t try criticizing Amazon in front of a self-published author. They would not be in business without The Everything Store.

To gain a perspective on AI’s coming impact on book publishing I recommend Clayton Christensen’s The Innovator’s Dilemma: When New Technologies Cause Great Firms to Fail, first published in 1997 by Harvard Business Review Press.

Christensen looks at how existing (incumbent) firms can succumb to the forces of innovation.

Successful, well-managed companies often fail when disruptive changes come to their industry. Conventional management practices, which had helped them to become industry leaders, make it difficult for these companies to be nimble in confronting the disruptive technologies that could cannibalize their markets.

They ignore the products spawned by disruptive technologies, because, at first blush, they compare poorly to their existing products. Their most profitable customers generally can’t use the purported innovations and don’t want them. Companies try to fight off disruptive threats by doubling down on existing products and services.

Christensen’s key insight is that by doing the apparently ‘right’ thing, including listening to customers, successful companies leave themselves wide-open to disruptive innovation. They focus on their current customers and ignore important new technologies—which initially target small, less profitable markets. It creates an opening for agile startups to disrupt the industry leaders.

Chris Dixon, in discussing the book, points out that “the reason big new things sneak by incumbents is that the next big thing always starts out being dismissed as a ‘toy.’” Hmm, yes, a lot of people see Chat AI as a toy.

Are publishers failing to recognize that innovations built with generative AI might disrupt their fragile business model?

AI’s impact on book publishing will have a dramatically different impact on the authoring and publishing of nonfiction than it will on the authoring and publishing of fiction.

The book publishing industry isn’t often enough analyzed as two bifurcated industries, one that publishes fiction books and another than publishes nonfiction.

While most trade publishers offer both fiction and nonfiction titles, the industry-wide gulf between the two forms is stark both in the number of titles published, and in book sales. Fiction titles represent roughly 10% of books published each year. Sales, however, greatly favor fiction. There are year-to-year shifts, but fiction captures up to half of annual trade book sales. In 2024, none of the top 10 bestsellers were nonfiction. Using Wikipedia’s compilation, 84% of the books in English that have ever sold 20 million copies or more are fiction titles.

Book sales trends are notoriously fickle, but fiction sales appear to be on a continuing upward slope. After cratering at 32% in 2019, they claimed 40% of the adult market in 2022.

While AI is going to be increasingly used by fiction authors to aid in the process of creation, as discussed elsewhere, there seems little likelihood that AI-generated stories will take over the bestseller shelves. “Good-enough” AI-generated genre fiction is conceivable, but again not something that should give authors and publishers sleepless nights.

The writing and publishing of nonfiction, on the other hand, will be aided and abetted by AI across the board. It’s happening already. Nonfiction authors are taking advantage of Chat AI’s many talents both as a research assistant and a writing aid. And nonfiction publishers will increasingly call upon Chat AI for help in manuscript development, fact-checking, editing, marketing and distribution.

When a reader approaches a work of nonfiction the book has no idea who they are. Perhaps they’re a high school dropout; perhaps a PhD. They might be brand new to the subject, or maybe just boning up. The “killer app” for nonfiction publishing dwells near here. Chat AI can customize text to your particular needs and deliver it rapidly, as an ebook or an audiobook, in 200 languages. Print does not compete. Print does not compute.

I spoke to a very smart author’s agent, who has been thinking a lot about what AI means to her clients, and to all professional authors. The conversation turned to the idea of the book as a ‘container,’ and I asked her to expand on that idea. Her response:

“The basic idea is that when authors/agents are placing a book with a publisher, it is just that: a book. There is an understanding that it might later become one of the myriad forms of derivative works we have come to know (translation, dramatic adaptation, graphic novel, etc.) but all of that is secondary to the ‘the Work’ itself. The Work is not a bag of words or phrases or facts, it is something structured by the author in a way unique to that author. Breaking the container, shaking the Scrabble bag of letters and drawing out a random selection (that is not entirely random because it is a Scrabble bag made up of the syntax and semantics and style of the specific author), is not what is being considered when we make a grant of rights to a publisher.

“The book is the integral whole that is greater than the sum of its parts. It’s the product of a creative epiphany (and a ton of creative intellectual work) that brought it all together as a book, not just ‘a’ book, but the book by that particular author.”

When I think about nonfiction books, breaking the container is one of the upsides of the AI era. You break the container, and like Fabergé eggs, there are treasures contained within.

The constraints of the container are both a feature and a drawback. On the one hand “the work is not just a bag of words or phrases or facts, it is something structured by the author in a way unique to that author.” On the other hand, in creating the container, the author, by dint of the inherent limitations of the bookish container, was forced to reduce, reject, rewrite, and reconfigure. The work is polished, but it is distilled. Some things were lost (while others were gained) along the way.

People fail to read long nonfiction books all the way through, in part because they’ve become conditioned to skimming all the text they encounter on the web—even in their emails. The “book summary” model fails for the same reason—instead of an indigestible 250-page nonfiction book you get an indigestible 12-page summary of the book.

Books are like lunch boxes—all the stuff in one place. But I just want the cookies.

It’s not going to be easy for established authors to throw off their chains. But they may have to learn to do so, or risk being overshadowed by a new generation of skilled researchers and writers who can express themselves with or without formal containers, as the occasion calls for. Rigid containers impose far too many limits, and limitations.

What it comes down to is that the business of writing has irredeemably changed. It’s like the end of the scribes. In this case, our monkish authors must get out from behind their desks and their containerized view of the book, and engage with form and with their audiences.

Oh yes, I know well the author rejoinders: But I’m a writer, I’m not a promoter. I’m not on social media. I don’t have a platform. I speak to my readers via occasional interviews and ever-more-occasional appearances. I have a website (but, truth be told, I never update it). My work is my gift to readers. They can accept my work, or reject my work, but I will not be joining them in their living rooms.

Well, you can put away the scrolls and the ink made from the blood of ermines. The occupation of ‘author’ now demands that you get out from behind your desk and meet your readers where they live, in their living rooms, and on their Facebook pages. You can lament it all you want, as we curtail your grants and your tenure and your publishing opportunities. This is the brave new world of writers and readers, sans publishers, those dreadful intermediaries who work hard to enhance and to market your work, yet constantly interrupt the direct connection to your readers, the people who most value your work.

None of this precludes great narrative nonfiction books reaching an attuned audience. Au contraire. Those books can continue to shine. Look at the recent Pulitzer Prize nonfiction winners. No one wants publishing to lose its ability to incubate books of this exceptional quality. That would be tragedy. The current publishing business model doesn’t need to disappear for the new model to thrive.

I think of these great nonfiction books as ‘beautiful.’ ‘Beautiful’ as in beautifully-written, carefully edited, designed with care and deliberation, and printed, with deckled edges, on FSC-certified paper. They are sold in boutiques.

The boutiques are called ‘bookstores,’ but they have all the characteristics of boutique retail—lovely objects, chosen with care by the owner and the managers, a bit pricey, but for those who treasure such objects, well worth the cost.

The task of most nonfiction authors is now to move beyond the container, as well as the publisher as gatekeeper.

The reading public, particularly its younger members, is increasingly ‘media-agnostic.’ In search of entertainment or knowledge, they can be watching a YouTube or TikTok video, a Netflix series, listening on Spotify, scrolling through Instagram (“Insta”), catching up on the news, or, occasionally, reading a book.

Yet the business models are still focused on the unique container. The publishing industry. The film industry. The television industry. The music industry.

Movies are not books. YouTube videos are not movies. Podcasts are not audiobooks.

These silos were not an historical inevitability but a result of a series of technology and business decisions made by powerful organizations over time.

It’s not hard to disrupt a business model built upon a single container.

For all of its facility with language, the current generation of generative AI works its magic with even more bravado producing sounds, images and video. AI affords this book with audiobooks, in multiple languages, alongside the 31 translations of the ebook. None of this would be even remotely possible without AI—the economics of traditional production makes no sense for me.

Content creators are traditionally trained to excel in just a single container silo. We have writing programs and film programs and music programs. It rarely occurs to us that creative writers might also take a ‘minor’ in film or music. Or in computer programming or in app development.

Whatever happened to transmedia? Where are the programs that teach creatives to be, as Apple once framed it, “Masters of Digital Media”?

The best content will always win. But, over time, the content containers matter less and less. The digital generation may not be entirely container-agnostic. But they are container-flexible. They encounter most of their content digitally, via smartphones. Print will continue to find an audience, but it will never again command the center.

The “publisher of the future” needs also to be container-flexible. Most of the growth in content distribution is not via ink on paper, nor words in EPUB files. The growth of audiobooks is no accident, nor is the popularity of YouTube and TikTok.

The larger challenge lies in the necessity of making new and existing content discoverable, whatever its form, and then upon discovery, to turn awareness into a purchase decision.

The online discovery problem has been with us for some time; AI did not create it. The problem is one that industry veteran Mike Shatzkin highlighted more than once. There are so many titles available in print, and digital formats allow them to remain continuously in print. Millions of out of print books can be found from used book retailers. On top of that two million or more new books are published each year, just in English. AI will make the saturation problem worse, but it’s already out of hand.

Amazon’s algorithms are not selfless—they do a very good job of surfacing what’s saleable, the book that you’re most likely to purchase next. That book isn’t characterized merely by its sales figures. It sold best not because some scammer optimized the online listing. It sold best because it was the best, and delighted readers told others to buy it, both online and in person. (Amazon encourages advertising that distort its algorithms, part of the enshittification of the platform.)

I talked above about metadata. Conversion is as vexing as discovery. Can AI help authors turn browsers into buyers?

The concept of copyright has been fundamentally challenged by developments in artificial intelligence. It’s not that authors and other creatives don’t want or deserve protection for their work—arguably they deserve it more than ever. And it’s not that AI renders copyright unenforceable (at least in some form, if not in the current form).

It’s that the “protection of copyright” can also become a veil behind which your work disappears. If your book cannot be at least referenced via a conversation with Chat AI then it, in essence, does not exist on that platform. Why bother protecting that which cannot be found?

This is where the discovery problem might become a catastrophe. Google does a great job discovering books only through their metadata. Omnivorous AI prefers the whole enchilada. Unlike traditional search engines, AI rarely responds to user queries based only on the abstract metadata about a work.

At the same time, evolving content containers, containers other than whole books, suggest different kinds of copyright challenges. If the text of a book can change on the fly, in response to reader input, what then is the copyright of the text? And how can the original author be fairly compensated?

Publishers need authors; authors don’t need publishers.

The future of publishing is the intimate relationship between writers & readers. It’s far stronger than the relationship between writers and publishers and readers and publishers. Publishers can be a roadblock in the relationship between writers & readers. In many cases they do not enable the relationship; they impede it.

For a long time, the only way to access high-quality written content was through books or via a modest selection of periodicals. That’s certainly no longer the case. Quite apart from the distractions of other media, there are now so many different ways to access (non-containerized) high-quality written content. Books no longer hold the primacy they once enjoyed.

“What’s the perfect book for me to read next?” Generative AI can answer that question with an eloquence and a precision that has never before been possible. As AI becomes a reliable recommender engine, authors will need to communicate via that engine. (Amazon, of course, is working on this, and has the benefit of knowing what you’ve previously purchased.)