On September 27, 2023, Publishers Weekly held a virtual half-day event, “Artificial Intelligence: Revolution and Opportunity in Trade Publishing.” The event offered a deep dive into the potential of AI within the publishing industry. The session “AI and the Business of Publishing” brought together experts in the field to discuss the challenges and opportunities AI presents to the world of publishing.

The panel, moderated by Thad McIlroy, digital publishing analyst and Publishers Weekly contributor, included Barbara Kline Pope, executive director at Johns Hopkins University Press; Michael Bhaskar, writer, researcher, and publisher; and Catherine Weldon, AI and machine learning associate at Penguin Random House.

The session commenced with a recognition of the legal landscape and the transformative role that AI plays in the world of publishing. The discussions revolved around the challenges posed by AI and the evolving business dynamics.

Bhaskar brought a unique perspective to the conversation. He highlighted that he has experience working both in the world of AI, having collaborated with Google’s DeepMind, and in publishing as the founder of a startup called Anello. Bhaskar emphasized that the intersection of AI and publishing is a pivotal moment. He also posited that AI’s impact on publishing is more profound than the advent of e-books, as AI is fundamentally changing the way human intelligence interacts with language and content. He emphasized the seismic shift brought about by AI, where even the core aspects of publishing are becoming susceptible to automation. This transformation presents a significant challenge and opportunity for the publishing industry.

Weldon provided insights into her role at Penguin Random House. She emphasized that AI is not entirely new to publishing and discussed her focus on research and development. She highlighted that AI is playing an increasingly significant role in various areas of the publishing business, from recommendation engines to forecasting models.

Kline Pope addressed the implications of AI for large university presses. She acknowledged that AI is creating new opportunities for increased productivity but also highlighted the need for guidelines and guardrails to ensure responsible and ethical AI implementation. Pope mentioned the importance of fostering creativity and innovation while ensuring that AI complements human skills rather than replacing them.

The conversation then delved into how the responsibilities of managers in publishing differ from those working on the front lines of AI implementation. Kline Pope underscored the need for leaders to empower their staff to use AI to automate mundane tasks, thereby freeing up their time for more creative work. She also emphasized the importance of instilling confidence in staff to experiment with AI and the necessity of providing clear guidelines for responsible AI use.

Bhaskar echoed the significance of encouraging experimentation and innovation within publishing organizations. He highlighted that management’s role involves not only fostering experimentation but also making judgment calls about the appropriate use of AI and guiding the long-term strategy to adapt to the changing landscape of publishing.

Weldon discussed the interactions between her role as an AI and machine learning associate and staff at Penguin Random House. She mentioned the increased focus on education, communication, and collaboration within the organization, driven by the adoption of AI. Weldon also touched on the importance of setting guidelines and maintaining clear communication channels to ensure responsible AI use across the company.

The conversation then shifted toward the potential downsides and risks associated with AI in publishing. Weldon expressed concerns about the potential inaccuracies in AI-generated content and highlighted the importance of verifying and proofreading AI-generated materials to maintain quality and accuracy. Bhaskar discussed the challenges of copyright and licensing when dealing with AI-generated content. He emphasized that there is currently no established model for monetizing AI-generated content and that it poses a significant challenge to the intellectual property world, as the contribution of a single book to a massive AI model is minuscule.

Kline Pope raised concerns about AI’s potential to amplify biases and structural racism in society. She stressed the need to address this issue as AI models continue to learn from existing biases and potentially perpetuate them on a larger scale. The discussion also touched on the challenges of managing relationships with authors in an evolving publishing landscape. Kline Pope mentioned that her press has received requests to include language in contracts that ensure author consent for AI use in their works. This reflects the need for a delicate balance between protecting authors’ intellectual property and embracing new technological advances.

The session ended by considering the practical applications of AI in publishing. Bhaskar highlighted the potential of AI in optimizing processes, enhancing marketing efforts, and improving content discovery. The speakers agreed that AI is being utilized to improve various aspects of publishing, from demand forecasting to the creation of marketing materials and book club guides. Overall, the session provided valuable insights into the transformative potential of AI in the publishing industry. While it offers numerous opportunities for streamlining processes and creating innovative content, the challenges and risks associated with AI, including issues related to copyright, bias, and accuracy, need to be addressed. This dynamic interplay between technological advancement and responsible implementation presents both excitement and uncertainty for the future of publishing. As the publishing industry adapts to this AI revolution, it will need to strike a balance between embracing innovation and upholding the core values of quality and integrity in literature.

Thad McIlroy on the Significance of AI

Thad McIlroy acted as co-host of Publishers Weekly’s “Artificial Intelligence: Revolution and Opportunity in Trade Publishing” virtual conference. We caught up with him after the event to discuss his thoughts on AI.

You mentioned during the event that AI feels different from previous technological revolutions. Could you elaborate on what sets AI apart in the context of the publishing industry, and why it feels significant to you?

Everything we’ve done up to now with technology in book publishing has involved doing the same old thing with better tools. AI changes both the inputs and the outputs of the publishing process. That’s fundamentally disruptive.

The typical response to new technology is caution, followed by gradual adoption. From your perspective, do you see AI as a transformative force in the publishing industry, and if so, how do you envision it reshaping the industry in the long term?

AI is transformative, in part because it is about language, and also because it touches on virtually every aspect of the book publishing process. I do think it will change the way that content is created, although I wonder if that will be its most powerful impact. From my perspective as a consultant to publishers, focused on technology, I see AI providing tools that will make publishers far more efficient and far more effective. At the same time, these tools increase the ability of “outsiders,” like self-published authors, to recreate, at a very high level, most of the functionality of the best in the business. It accelerates a power shift that we have been seeing in publishing since the dawn of digitization.

In your introduction to the event, you highlighted the importance of generative AI and its ability to emulate reasoning based on language. How do you see this capability impacting the creation and understanding of content in the publishing industry?

This question is hugely complex. AI can be used both to create new content and to help us understand existing content. Understanding existing content is easier to visualize—an obvious example is that we can use AI to give us intelligent summaries of long documents, reports, emails, and websites. People like to make fun of AI’s content creation capabilities. That’s understandable, because early users have engaged lightly with AI-generated content, and the first-pass results are often silly and superficial. We get into arguments about whether AI can be “creative” in generating content. I point out that most humans generate quite ordinary content—emails, documents, reports, and so on—and that serves us perfectly well. Just because AI won’t become the next Gabriel Garcia Marquez doesn’t mean that it can’t produce a slew of material that can educate and entertain.

The event covered using AI in editorial, marketing, and production functions. What do you think are the best uses of AI today, and do you think that may change?

The easiest way to bring GPT into an organization is through marketing, followed by editorial. Production is more complex. Over time, I think that generative AI technologies will penetrate deeply into all publishing functions.

What are some practical challenges or obstacles that publishers may encounter when implementing AI technologies, and how can they overcome these challenges?

As with all new technologies, ignorance and fear are the largest obstacles. With knowledge you can overcome fear (or, alternately, decide that your fears are well-placed, and abandon that technology). How to  become knowledgeable? The tools are cheap and relatively easy to use, and so there is not a significant initial barrier to access. But understanding their utility at a more profound level does require both experience and learning. Large publishers can acquire some of that by hiring. Smaller publishers have to bite the bullet and commit to training their staff.

It’s evident that there’s a lot of AI-related activity taking place, especially in startups. Do you have any insights into the specific areas within the publishing industry where startups are making significant strides with AI, and how might these startups shape the future of publishing?

I’ve tracked over 100 book publishing-oriented startups with AI as a key component of their product offering. Tellingly, well over half of the startups target authors. Book readers are the second largest group, followed by book publishers. What this points to is that the innovations around AI are likely to start in within the self-publishing community.

What are some key AI tools or solutions that you believe publishers should be aware of?

The core technology is available online for little or no cost: OpenAI’s ChatGPT (which can also be accessed via Microsoft tools, such as Bing), DALL-E, Google Bard, and Anthropic’s Claude. Those are a good starting place.

What advice would you offer to independent publishers looking to navigate this rapidly evolving landscape?

As with most technologies, the smaller the organization, the tougher it can be—smaller publishers don’t have the technical resources on staff and don’t have the cash to hire tech consultants. The good news about AI is that, while the underlying technology is extremely complex, you really don’t need to understand any of that to engage with the software. Just start playing with ChatGPT and see where that takes you.

---