Professionals in the book publishing industry are terrified of artificial intelligence — and for good reason. While it isn’t all bad, it has the potential to bring the sector to its knees. Not to mention, it has already harmed many creatives. Are their fears justified or unwarranted?
Thanks to its versatility, AI has emerged in most parts of the publishing industry. In 2023, 47% of book publishersbased in the United States used it for marketing, 25% for editorial work, and 12% for content generation. Companies can use it in writing, editing, advertising, or administrative roles — it can do almost everyone’s job to some extent.
Many disruptors have affected the publishing industry. Most recently, e-books have started replacing physical books. According to one estimate, about 30% of pages will be digitally printed by 2028, up from 5% in 2023. Many publishing houses and self-publishers are migrating to digital. Before that, digital printing shook up the offset printing press.
Although AI isn’t the first disruptor, it’s the biggest so far. The publishing sector has withstood others because they were all natural evolutions to outdated processes. Machine learning (ML) and large language models are different — they replace existing technology instead of improving it. Who needs writers, editors, and agents when algorithms can do their work for free?
Most professionals in book publishing are terrified of generative AI. Its versatility could bring about the sector’s downfall. A handful of high-quality models can write a novel, proofread, edit, design a cover, make marketing materials, post on social media for publicity, distribute digital copies, and handle accounting.
For now, AI can’t perform any of these jobs better than a human. One industry professional said the content it generates is “crude at best,” arguing it isn’t comparable to what a real writer can produce. Even so, it might not matter when it can pump out pages at an inhuman speed.
AI will inevitably affect the entire industry even if every author, editor, and publisher agrees not to use it. This technology’s accessibility is its main driving force. Dozens of leading models — and thousands more in-house systems — exist. Almost anyone can use ChatGPT, Midjourney, LaMDA, or Claude to make, market, and distribute AI-generated books.
There’s abundant evidence of AI’s negative impacts on the book publishing industry.
There’s truth to the fear that AI is set to steal people’s jobs. While more job ads now stress human-written content is a hard requirement and more companies are investing in software to detect AI-generated content, the publishing sector’s pushback hasn’t been enough — and people are being forced to find new careers.
Olivia Lipkin, a copywriter, said her assignments dwindled after ChatGPT came out. She even claimed management changed her title to “Olivia/ChatGPT” on the company’s messaging platform. Within a few months of the generative model’s release, Lipkin was let go without explanation. Eventually, she discovered her managers had realized AI was much cheaper.
Algorithms can’t come up with content on their own — they must use others’ work. Since most books are copyrighted, AI companies have resorted to piracy and copyright infringement. Books3 is a data set of nearly 200,000 pirated novels used to train generative AI systems. Most of the content was published within the last 20 years, so many living authors are affected.
This issue isn’t limited to bad actors and shady characters, either. OpenAI, ChatGPT’s creator, argued creating models would be impossible unless it was allowed to use copyrighted material. It claimed public domain content wouldn’t “meet the needs” of people. Regardless of the statement’s truth, it shows how unconcerned AI companies are about content theft.
Despite the issue of AI-driven copyright infringement being a real threat to authors, literary agents and publishing houses, no real regulations exist to protect them yet. Recently, a 24-page document revealed that Midjourney had a database of around 16,000 individuals it allegedly stole art from. Apparently, it was a list for model training.
Even though the list — which contained specific names — seems like it would prompt swift regulatory action, things aren’t as open and shut as they should be. In many countries, a lawsuit is the only recourse creators have. Realistically, there’s no guarantee they’ll be successful. Not to mention, many would have to fight against well-funded tech giants.
AI wouldn’t be able to disrupt the book publishing industry if it was inferior to what people currently use. In other words, a silver lining does exist — there are potential positive impacts.
Most writers spend more time than they’d like on mundane — but necessary — tasks like proofreading and spellchecking. The same can be said about literary agents hassling publishers or marketers tweaking designs for the nth time. AI can take over these boring, repetitive duties.
Since ML models evolve over time, they can create and maintain searchable databases of relevant marketing avenues, contract work or agents to query. Publishing companies can leverage AI tools to help editors check for grammar or continuity mistakes, accelerating time to market.
Publishing AI-generated books is a get-rich-quick scheme. Natural language processing technology isn’t advanced yet — readers and literary agents can quickly identify this kind of content. Fortunately, many view it as cheap, poor quality or lazy.
To avoid AI-generated books, readers must search for authors. In doing so, they’ll spotlight creators and learn to appreciate content better. Many companies are already taking advantage of this fact, using “100% human-generated content” as a selling point.
AI’s impact isn’t purely black and white — there’s a massive gray area where things could go either way. Navigating these subtleties is tricky.
Book publishing is notoriously difficult. Even now-famous authors have dealt with round after round of rejections. An ML model in a decision-making role could evaluate content at an unparalleled pace, helping more people get published.
It may occasionally make biased or illogical decisions, but those aren’t exactly uncommon. Numerous cases exist where a book goes through dozens of rejections before becoming a best seller — well-trained literary agents and publishers make mistakes, too.
Anybody with access to AI can ask a model to write a book, design a cover, and generate hype on social media. This technology makes self-publishing easier than ever, meaning previously overlooked and underserved groups can fulfill their dream of publishing a novel.
Unfortunately, bad actors are taking advantage of the simplicity of AI-driven self-publishing. One author, Jane Friedman, found six fake titles under her name in a few months. Others claim they spend hours reporting listings just for new ones to pop up the next week.
Another potential issue with AI-powered self-publishing is that authors won’t need professional help anymore. If most people go this route — which they might because getting published traditionally can be next to impossible — the sector could collapse.
Although every big disruptor causes general worry, AI is different. For one, it’s accelerating at a pace publishers can’t keep up with. It’s also like no technology people have ever seen before — what other invention can autonomously read, learn, write, edit and publish a novel? Professionals’ fear seems to be justified. For now, there’s no telling if generative models will transform, replace or topple this industry.
This article was originally published by Zac Amos on HackerNoon.
Expanding into new geographic locations is one of the most effective ways to drive growth…
Every now and then, I stumble upon posts such as these here and there: And,…
Winter(Physics) is Coming It now looks like Large Language Models running on the GPT technology…
Latin America’s tech industry is booming, with innovative new startups popping up across the region.…
The Global Initiative for Information Integrity on Climate Change claims to 'safeguard those reporting on…
In the late 19th Century, physicians began inserting hollow tubes equipped with small lights into…