Like it or hate it, generative AI tools have started to visibly affect book publishing processes—editorial, design, marketing, and more. Book designer Damonza recently posted their AI policy (with a discount for clients who allow AI to be used in their designs), while traditional publishers have been blindsided by the discovery that their own covers have, unbeknownst to them, included AI elements. It’s hard to even keep up with the evolution of these tools; new developments (and legal actions) are announced daily.
Earlier this month, the Women’s Media Group convened a panel to discuss emerging AI technology and how it’s affecting authors, agents, and publishers. Participants included Cecilia de la Campa of Writers House (a literary agency); Catherine Weldon, an AI and machine learning scientist at Penguin Random House, whose job is to develop predictive and generative AI solutions to enhance publishing and marketing efforts; and Amy Willstatter, an executive with Charisma Entertainment, which has developed an interactive conversational AI platform for storytellers. Moderating the discussion was Charlie Schroder, head of original content development and publishing at subscription service Scribd.
Agents are trying to ensure author contracts encompass all eventualities of this technology. Ideally, any use of AI hinges on the author’s approval, so that each opportunity can be assessed as it comes up. De la Campa says that’s “the best mutually beneficial way to enter into new business but also have some protections.” She says the four big areas they focus on at Writers House are translations (which she believes are going to boom with AI), audio and voice, language model learning, and cover design. If authors don’t want their books translated by AI or turned into audio by a synthetic voice, that should be in the publishing contract. She says the easiest area to negotiate right now is covers. Even though covers are the publisher’s responsibility, if readers discover AI elements have been used in the cover, the blowback typically comes down on the author. “It’s a problem for the author and their brand,” de la Campa said. She said authors also need to be careful when working with website designers, marketers, and publicists who might be using AI-generated imagery or materials. Outline your expectations and if there are any circumstances where you’re okay with AI being used.
Penguin Random House is using generative AI to support marketing efforts, particularly book summaries. However, Weldon said, there’s risk involved because the tools will generate misinformation. She has not read every book that PRH has published and doesn’t know if the summaries generated are entirely accurate, even if they sound plausible. So she thinks there will always need to be a human who really knows the book to review AI-generated material. It’s still a time-saver, though, to automatically generate some of these elements, then review and edit them.
Those worried about PRH loading entire manuscripts into ChatGPT can be reassured that is not happening. PRH can host open-source versions of these tools internally, so that book content isn’t openly accessible or feeding the AI learning models. However, Weldon says, “It takes a lot of resources to train these models. PRH is not investing millions of dollars to have our own model. But we are fine-tuning these models to make them more specific to our own purposes.” This has the added benefit of making the output more reliable.
Generative AI can have positive outcomes: The first is accessibility of materials. AI can make it easy and cost-effective for all books to be accessible to more communities, particularly disabled communities. AI can also improve productivity by allowing editors and marketers to save their time for higher-level and creative responsibilities instead of generating marketing keywords and summaries. Of course, some are worried that any time savings will lead companies to reduce staff, but this has always been the fear surrounding new technology. Usually it’s not the fault of technology; it’s the fault of executives trying to please shareholders, using tech as a scapegoat. Weldon said, “I hope to see legislation move a bit quicker and create safety nets, so that this increased productivity, which could be a really great thing, is evenly distributed and people can still do what they love, create art, and write, and have dignity.”
Bottom line: No one disputes the risks (such as the generation of misinformation or potential infringement) involved in using generative AI. Just as everyone had to learn how to run good internet searches and develop online literacy, they will need to learn responsible use of generative AI. Weldon says the models will become easier to train and cheaper to train over time—meaning businesses and even individuals will be able to develop small, custom models for their own purposes.
Jane Friedman has spent her entire career working in the publishing industry, with a focus on business reporting and author education. Established in 2015, her newsletter The Bottom Line provides nuanced market intelligence to thousands of authors and industry professionals; in 2023, she was named Publishing Commentator of the Year by Digital Book World.
Jane’s expertise regularly features in major media outlets such as The New York Times, The Atlantic, NPR, The Today Show, Wired, The Guardian, Fox News, and BBC. Her book, The Business of Being a Writer, Second Edition (The University of Chicago Press), is used as a classroom text by many writing and publishing degree programs. She reaches thousands through speaking engagements and workshops at diverse venues worldwide, including NYU’s Advanced Publishing Institute, Frankfurt Book Fair, and numerous MFA programs.
