Like it or not, artificial intelligence is here to stay, and it is causing seismic waves within the publishing industry. How can publishers grapple with the ever-accelerating pace of change, the potential for streamlining business operations, and the ethical dilemmas that come with AI?

The Quandary

“AI is not a passing fad,” says Emerald Lake Books publisher Tara Alemany. “It’s a skill set. And any business employing people who have this skill set is finding they can do more with less and quicker. That’s hard to ignore in an industry with traditionally tight margins.”

Alemany notes a change in how publishers approach AI. “Two years ago, many publishers were either experimenting quietly or avoiding AI entirely,” she says. “Today, I see more openness to practical, behind-the-scenes uses of AI, especially for operational efficiencies and marketing.”

She attributes this shift to the “sheer volume of real-world examples” showing how publishers use AI to improve their business. “Publishers can see what’s possible, what works, and what doesn’t,” she says. “That makes the conversation less theoretical and more grounded in results.” 

But ethical concerns, along with legitimate fears, remain part of the conversation. “While I am a proponent of everyone working in publishing becoming highly AI literate, I think AI poses an undeniable threat to the fabric of our society,” says Perfect Bound publisher and Stable Book Group President Keith Riegert. “We are already seeing mass layoffs across myriad industries being driven by AI automation, and we’re still in the very early years.”

Years ago, Riegert recalls watching a documentary about climate change with his father. His father pointed out that if the predictions were accurate, humankind would come to view the Industrial Revolution as the biggest mistake in human history.

“I’ve been thinking about that moment a lot over the past couple of years,” Riegert says. “I think AI is going to be a double-edged sword. I think it will allow for incredible discoveries, like cures for cancer [and] maybe even solutions for climate change. But we are intentionally ceding our position as the most intelligent and creative beings on this planet to something synthetic that a handful of people have engineered. Entire professions are being automated, and we are already seeing a flood of AI-generated content compete with human-made books on Amazon.”

Even so, Riegert says companies that are embracing AI now have the potential to become “very profitable.” Before long, AI will become the norm. “There is no doubt in my mind that every industry, including publishing, is going to become very reliant on generative AI, AI agents, and AI workforces,” he says. 

Publishers may be more open to AI than they were a few years ago, but Riegert says adoption has nevertheless been slow. “I’ve continued to see a lot of resistance to even trying AI across every employment level in publishing,” he says.

Thad McIlroy, president of The Future of Publishing and author of The AI Revolution in Book Publishing, urges publishers not to panic over AI, but to plan and move fast. At the same time, he acknowledges publishers’ concerns about copyright, market and workforce disruption, and the potentially negative impact of AI on society and culture, especially creativity.

McIlroy calls this the “weird little witch hunt phase” of AI adoption in publishing. Within that phase, he sees author-publishers as trailblazers. “With AI, it’s giving author-publishers a huge advantage over tradition-bound publishers,” he says. “They’re fearless. They’re far more open to new technologies.”

Publishing is a conservative industry, McIlroy says, and AI is changing the landscape quickly. “Most people underappreciate how good the tools have gotten—frighteningly good,” he says. At the same time, ethical concerns abound. “It’s an ethical minefield,” he says.

But if publishers only tiptoe around that minefield, they may find themselves on an unsustainable path. “The tools of our trade are changing,” Alemany says. “If we don’t change along with them, we’re only making success harder to reach.” 

Streamlining Operations

Efficiency is vital to success, and AI offers myriad affordable ways to achieve it. But what about the human cost?

For Riegert, using AI isn’t about replacing jobs but freeing up employee time with the automation of redundant tasks. “The best thing about working in publishing is the opportunity to create,” he says. “My hope is that we can use AI as an industry to foster creativity and innovation by passing off time-consuming tasks like royalty accounting and manual marketing optimization to AI.”

Alemany cites several ways AI has helped her streamline operations. “As a solo business owner, ChatGPT has become an invaluable resource to help me accomplish a lot more in less time with a higher degree of confidence,” she says. She uses Descript for shooting scripted videos and Perplexity for fact-checking. She is also experimenting with Storywise, an AI system that does a quick, rough assessment of submissions, allowing publishers to prioritize the most promising projects.

Alemany also lauds the potential of AI’s application programming interface (API), which streamlines operations by connecting systems. “API-driven automation has been a quiet but powerful game-changer,” she says. “Tasks that used to take 20 to 30 minutes each, like generating preliminary keyword lists, performing research, assigning tasks to my team, and evaluating potential new tools, can now be triggered automatically by specific events.”

Specifically, Alemany uses Pabbly Connect to integrate AI into her publishing workflows. “When a new author submission comes in from our website, Asana tasks are created and assigned using templates that already describe what needs to be done,” she says. “A folder is created in Dropbox under the prospective author’s name and a PDF copy of their submission is automatically saved into that folder, all without human intervention.”

Likewise, when a customer makes a purchase on Emerald Lake’s online store, AI generates a sales receipt in QuickBooks, sets the status of the order to “processing” in Alemany’s Ecwid store, adds details of the sale to a Google Sheet that she imports into her royalty tracking system, and creates an Asana task to ensure timely shipment.

“For publishers, the benefits of things like this are twofold,” Alemany says. “You free up staff to focus on high-value work and to put a human face to your business, and you reduce the lag time between steps in a production or marketing process. That means things can move faster, and you can spend more time working on the business than in it.”

Wherever possible, Alemany trains AI tools to align with her company’s unique brand and needs. “Overall, the big shift for Emerald Lake Books has been moving from asking AI for one-off outputs to training it in our processes so it produces consistent, on-brand results,” she says. “This allows me to use it more effectively as an additional contributing member of my team.”

With ChatGPT trained in her brand voice, Alemany uses it to build structured templates for EPUB accessibility assessments, create AI-powered workflows for refining the company’s alt text, generate market-specific promotional calendars, develop strategic ad copy in the company’s brand voice, and generate job descriptions and requirements to post to Upwork. “We have even used it to build out cover design briefs after providing AI with a transcript from a recorded cover interview with the author,” she says.

Alemany even uses AI as a business coach. “I have one chat thread where I instructed ChatGPT in the style and method of business coaching that appeals to me most,” she says. Now, she routinely checks in with her “coach” for help with strategic planning, refining advertising campaigns, automating business functions, and learning new information.

Like a human, AI makes errors. “I have to be mindful of critiquing any output it creates,” Alemany says. “But it gives me a great starting point when I’m trying to accomplish something new.”

Reigert’s companies use AI to generate contracts, craft policy, generate book outlines, augment stock imagery, scrape websites, and join data sets. The tools they use include Chat GPT, Photoshop, Veo 3, Midjourney, NotebookLM, Kling, and Intercom. By training a Custom GPT on five years of sales conference data, they’ve created a customized sales script generator and an A+ content generator for Amazon pages. At Riegert’s Ulysses Press website, an AI chatbot gives book recommendations, ordering information, submission guidelines, and rights instructions.

Riegert uses agentic AI tools like Relevance AI and Moveworks to help publishers spot trends, optimize digital marketing, and analyze dynamic pricing. In addition, he suggests publishers become familiar with generative engine optimization (GEO) for book discovery.

The Content Conundrum

The nightmare scenario of AI-generated books flooding the market is enough to steer some publishers away from AI entirely. But as with other matters involving AI, the situation is far from straightforward.

McIlroy points out that with nearly 50 million books in the market already, dumps of AI-written material are “small potatoes in the overall scheme of things.” He is also forthright about AI’s potential to create books. In commercial genres, “AI could generate something of comparable quality,” he says. “It does an alarmingly good job of emulating your style.”

McIlroy admits that “during these very sensitive times,” AI-generated content may be perceived as “akin to plagiarism.” Still, he wonders how much readers care about a machine’s involvement if the product meets their need, be it for information or entertainment.

Like many publishers, Riegert says he draws the line at AI creating publishing content, be it writing, images, or audio narration. “We’re very strict about that with our authors, editors, illustrators, and designers,” he says.

At the same time, he sees value in content creators using AI tools. “Personally, I feel that we should, as an industry, be ready to use, without stigma, AI tools for developmental editing, proofreading and formatting, and, for authors, research assistance, brainstorming, and outlining.”

Those who use AI for tasks such as editing face decisions about disclosure. “I’d advocate for full disclosure on the part of publishers, authors, and freelancers,” McIlroy says. “But there are going to be lots of negotiations in the years ahead on what that means. When what AI produces is superior, where does the argument go?”

Publishers grapple with what full disclosure means. “Sourcebooks says they won’t use [AI] for anything creative,” McIlroy says. “Is a good copy edit creative? Is a good index?”

Publishers may also face challenges of negotiating around AI. “There are authors who want an AI-free zone,” McIlroy says. “They’re willing to deal with the contradiction that they’re using AI in spellcheck and similar tools.” But McIlroy warns that limiting AI use with some projects and not others can force publishers to create two workflows.

McIlroy points out that while authors typically warrant their work as their own, publishers would do well to adopt policies requiring authors to be candid about their use of AI. “This is a transitional issue,” he says. “Ten years from now, no one will be talking about this.”

Policies covering AI use by freelancers can be even trickier. Hourly rates don’t make sense when freelancers use tools that produce results in minutes. At the same time, McIlroy says he wouldn’t work with an editor who doesn’t use AI tools (see sidebar).

Even if publishers set AI use limits on freelancers, how can they ensure the freelancers abide by them? “They can’t,” McIlroy says. “There are these so-called AI detection tools that are sometimes reasonably accurate, but they’re not reliable. [They] give you a false sense of confidence, and you’re going to end up accusing some people of using AI who didn’t use it.”

Emerald Lake uses AI tools for some editing and production tasks, but Alemany draws the line at AI being used to create book content. “When it is used to assist the author or publisher, such as brainstorming ideas, refining structure, or checking grammar, it’s no different from using a skilled editor, a research assistant, or even spellcheck,” Alemany says. “So, I don’t believe that requires disclosure.”

More broadly, Alemany frames disclosure decisions in terms of authorship and accountability. “If a human creator is still making the final creative decisions, then AI is simply a tool,” she says. “If AI is the primary creator, then I think it’s the publisher’s responsibility to be transparent about that.”

Copyright and Licensing

While cases involving copyright and AI make their way through the legal system, McIlroy notes that President Trump’s “America’s AI Action Plan” has the potential to be “game over” for copyright. “[Trump] is trying to make fair use the law of the land,” McIlroy says. “It’s important for publishers to think through what it means if it’s now legal for AI engines to take whatever they want.”

Whether an action plan can upend legal rulings and precedents remains to be seen. In the meantime, as publishers consider whether to license rights for AI training, Created by Humans CEO Trip Adler says the biggest hurdle is uncertainty. “Many publishers and authors are still working to agree on the basics of AI rights,” he says.

Publishers who want to explore AI licensing also face operational challenges. “Negotiating the right terms requires specialized knowledge of AI use cases, data handling, and content protection, which can be outside a publisher’s core expertise,” Adler says. “There is also a trust gap, as publishers want confidence that AI companies will honor the boundaries of a contract.”

Despite the hurdles, Adler says publishers are becoming more open to AI licensing. “Just a few years ago, AI was often viewed as toxic in publishing circles,” he says. “Now, more publishers are actively seeking opportunities to license their content, experimenting with pilot deals, and engaging in conversations about how AI can be an additive rather than extractive force in the industry.”

With a system that handles AI licensing logistics, Created by Humans finds that AI companies are offering better contract terms than they did at first, including limits on content reproduction and sourcing requirements. “These changes have helped build trust and open the door to more licensing,” Adler says.

Adler says he has seen “significant deals” from publishers who moved quickly into the AI licensing space. “Others are still in the exploratory phase of testing the waters with pilot projects,” he says. “The most important thing for the industry is that we are in communication with AI companies. We won’t succeed in isolation.”

Adler says there is heightened interest in licensing nonfiction catalogs, especially business, health, and STEM. “These categories lend themselves to AI applications that teach reasoning, answer complex questions, or power specialized agents,” he says. For retrieval augmented generation (RAG), which supports the retrieval and assimilation of new information for large language models, Adler says there is demand for everything from fiction to textbooks.

Strategic Adoption

Publishers who want to make the most of AI tools will benefit from a strategy that facilitates ethical and efficient uses. “Don’t just plug in tools,” says Erin Servais, who developed the AI for Editors course (see sidebar). “You need to redesign workflows around AI’s strengths and your bottlenecks. Have the machines do what the machines do best, and humans do what humans do best.”

Publishers should standardize AI use within their own companies, Servais says. “Too often I see publishers with teams that use a patchwork of AI tools, and no one is entirely sure who is using which tools and how,” she says. “That’s not just inefficient. It becomes dangerous if not everyone is using their individual suite of tools with the same level of discernment.”

As more entry-level employees arrive with post-secondary experience using AI, Riegert advocates for broad AI literacy across every level of a company. “Everyone needs to know what these tools are capable of and ensure organizations are operating with ever-adapting guidelines as the AI rapidly improves,” he says.

Hatch Endeavors owner Torrey Sharp consults with publishers on branding and AI solutions. “The tactical use of AI should follow a deep exploration of values and strategy,” he says. “At the very least, I would encourage publishing executives to learn all they can about AI, test use cases that lead to simple, efficient outcomes, and avoid being close-minded to the practical benefits of AI. Instead of thinking about AI in terms of relegating our creative output and thinking to a machine, consider how the machine can leverage our thinking and our creativity into previously impossible realms.” 

Torrey says an AI adoption strategy should begin with articulating how AI will advance the company’s vision and purpose. “Once a clear vision is presented, the next biggest challenge is isolating meaningful use cases that are easy to implement and evaluate,” he says. “Training is essential at this stage. Getting some early wins helps build consensus. Rewarding this discovery process will help cultivate an overall spirit of innovation within the ranks.” 

The time for publishers to be watching from the sidelines is over, Torrey says. “It’s not too late,” he says. “The tools have advanced significantly, becoming easier to adopt and more powerful. Learning at this stage is greatly accelerated.”

When incorporating AI, Alemany suggests publishers start small. “Pick one process that eats up too much of your time or budget and train AI to help with that,” she says. “Refine your prompts and templates until you get consistent, usable results. Once you’ve seen the return-on-investment in one area, tackle another.”

She recommends treating AI like an intern, not assuming it knows too much. “Ask it questions about what it knows and fill in the gaps so it understands you and your business,” she says. “Invite it to ask you questions about your business, too. I’ve explored some very interesting topics just trying to answer questions my AI coach has asked me about how and why I do things the way I do.”

Alemany also notes the importance of reviewing AI outputs, especially at first. “The more you work with AI in a structured and intentional way, the more it will adapt to your voice, standards, and goals,” she says.

As publishers incorporate AI into their workflow, do they risk becoming reliant on it? Alemany isn’t sure the question matters. “We’ve all become over-reliant on using a computer,” she says.

The AI Accelerator

For all the potential benefits AI offers publishers, ample concerns remain. “In the end, everyone’s personal choice is going to come down to merging with AI or rejecting it,” Riegert says.

Ultimately, McIlroy says the decision boils down to the same questions they’d ask of other proposed innovations: Is it faster? Is it cheaper? Is it better?

“The old chestnut is you can get two out of three,” he says. “With AI, you get all three. If the whole world shifts around us and we keep doing what we’re always doing, what does that mean?”

Alemany also points out that AI decisions don’t have to pit humans against machines. “When we use AI as an accelerator, not a replacement, we allow ourselves to focus on those aspects of ourselves that make us truly human,” she says.

AI for Editors

We asked AI for Editors (aiforeditors.com) founder Erin Servais to share her thoughts on AI and editorial functions.

When did you create the AI for Editors course? Why did you create it?

In Spring 2023, I tried using ChatGPT to copy edit for the first time and saw how, in about a minute, it copy edited a text on par with my skill level, only much, much faster. It was clear to me that this was the future of editing, and everyone would need to learn how to use it. I decided that day to wind down my small editorial agency. I poured all my time and energy into building the first version of the AI for Editors course, which I launched a few months later.

What reactions have you noted from publishers and editors about the editorial use of AI?

Overall, I’ve seen a mix of cautious curiosity, excitement about time savings, and concern about ethics and trust.

Traditional book publishers are seeing the utility of using AI to write metadata, descriptions, and marketing copy, along with pulling comp titles and doing some broad-stroke analyses. There was a fundamental breach of trust when AI companies used pirated manuscripts to build their models. Understandably, many publishers seem hesitant to use AI beyond this. There are examples of hybrid publishers and self-publishing companies using AI in a much more integrated way, such as helping ghostwriters to analyze interview transcripts and other source materials, all the way to editing text.

Editors still often begin their AI for Editors training skeptical that it can work well enough to help them. It doesn’t take long to illustrate AI's skills across the editorial workflow—developmental editing, line editing, copy editing, and fact-checking. But many share the lack of trust publishers have, and they’re concerned about how to use it ethically, including in consideration of environmental impacts.

What are the benefits of using AI for editorial functions? What are the drawbacks?

AI helps editors do their jobs both faster and better. At first, many were concerned they would be replaced by AI. While there were some companies that made some misguided decisions about this, by far, what I'm seeing is editorial teams being able to work through backlogs and be able to keep up with demand because they have increased efficiency. It also helps them to do better work because it's like having two editors in one. AI can serve as a second set of eyes to come up with fresh ideas, help unknot tangled-up sentences, and handle the rudimentary grammar, spelling, and punctuation changes so editors can focus on the higher-level work.

What would you say to publishers who want to screen freelancers based on their use of AI? If publishers set AI use limits on freelancers, how can they ensure the freelancers abide by them?

Instead of screening freelancers based on whether they ticked Yes or No regarding AI use, look at how they are using AI. Ideally, for data security, publishers will extend an enterprise license of their AI tool of choice even to their freelancers. It’s simply in their best interest to do so. Otherwise, ask questions to determine if the freelancers use paid versions of AI tools and set strict privacy controls in them. Test them to see if they verify AI output and grasp which tasks AI does well, like basic copy editing, and which tasks require advanced prompting skills, like line editing. Before they begin work, have them sign an agreement to follow your AI guidelines (which means you need an AI policy) and conduct ongoing spot checks to determine if they are following the guidelines. No one wants to become the next AI-related PR nightmare, so there must be accountability at every level.

How do you expect agentic AI to change the publishing landscape, especially as it affects the editorial workforce? When should publishers anticipate these changes? How can they best prepare and adapt?

For a year already, I’ve been teaching people to build AI agents that follow their custom style guide, meaning you input text and it automatically applies your style guide’s rules; that apply other rules such as plain language guidelines; that transform text, such as turning an article into social posts; and that consistently apply a writer’s or organization's voice, tone, and style. These types of agents are all net positives for publishers. It allows them to find more uses for the content they already have without adding to the workload. It improves consistency and quality. And it helps their teams operate more efficiently because they’re spending less time on necessary but time-intensive work.

---