On September 27th, 2023, Publisher’s Weekly held a half-day virtual conference on Artificial Intelligence: Revolution and Opportunity in Trade Publishing. With the debate surrounding the insurgence of AI into trade publishing, this was a welcome webinar for industry professionals trying to navigate the challenges that AI poses.
The conference was co-hosted by Thad McIlroy, digital publishing analyst and PW contributor, and Peter Brantley, director of online strategy for the University of California Davis Library and founder of the Books in Browsers conference. With opening remarks by the former CEO of Penguin Random House, Markus Dohle, the conference was structured into panel discussions not only touching on the big picture but also touching upon individual areas of publishing like authoring, editorial, marketing, and production. An overview of the legal landscape surrounding AI, Authoring and Publishing also formed a part of the four-and-a-half-hour-long conference. The conference wrapped up with keynote speaker Ethan Mollick, an Associate Professor at the Wharton School and concluding remarks offered by the co-hosts.
Co-hosts McIlroy and Brantley flagged off the conference with an overview of AI’s impact on publishing and offered a preview of the online seminar. First up, Markus Dohle shared his perception regarding the convergence of AI and book publishing. He considered AI to be the “next inflection point” in publishing, like the rise of e-commerce in the form of Amazon and the evolution of e-books. He pointed out the doom-and-gloom stories being circulated when e-commerce and e-books were being assimilated into the industry to affirm that just as printed books prevailed in the face of the previous digital adversaries, they would come out as victors again. Talking about the two main services (Creativity and Reach) of the publishing industry, Dohle cited the previous examples and put forth his belief that AI would only help enhance and optimize the industry’s reach further, making it even more personalized and reader-centric than ever. Addressing other concerns about AI, he said it was now the time to carefully navigate the legalities of copyright and regulate AI to protect authors and human creativity.
Following this, a panel comprising Barbara Kline Pope (Executive Director, John Hopkins University Press), Michael Bhaskar (Publisher & Author), and Catherine Weldon (AI and Machine Learning Associate, PRH) moderated by Thad McIlroy discussed the broader scene of the effect AI is having on functional areas of book publishing.
Michael Bhaskar set the tone for the discussion by building on Dohle’s point about e-books. He termed the rise of AI as the “earthquake moment” in the publishing industry as large language models are simulating human thought for the first time. In this experience, he has seen more of a fear and loathing of AI than acceptance of it and he believes that the intersection point will only arrive when the management has established strong guidelines for AI’s use. He looks forward to the moment when all media is one seamless, interchangeable whole.
Catherine Weldon seconded his opinion by pointing out the existence of AI before this moment but only on the statistical model side of things and that this was an important moment with a “bigger impact” because of the challenge posed by AI to the creative aspect of humanity. Her major concern was how the large language models churned out information so confidently that the impulse to use those answers as expert opinion needed to be checked. Fact-checking the information for inaccuracies should be a priority according to her.
Barbara Pope, who heads a staff of approximately 130 people at John Hopkins University Press, claimed to be very intrigued by the numerous possibilities of AI. She believed her press to be experimenting and being in the “toddler stage of AI,” but in her interactions with her team, she found that they preferred guidelines for working with AI. In framing these guardrails, one of the things that has become increasingly apparent is that at this nascent stage everything that is delivered by AI needs to be checked by a human first because of the biases creeping into AI. Her concerns lie with equity and how large language models could magnify the structural racism and bias that exists in society.
Next, panelists Stephen S. Power (Executive Editor, Kevin Anderson & Associates), Cliff Guren (Founder & CEO, Syntopical), and Barbara Ruehling (CEO, Book Sprints Ltd.) moderated by Peter Brantley considered how the editorial wing could use AI for their benefit. Brantley speculated that AI could be functionally used for summaries, fact-checking, plot descriptions, character descriptions and could also aid in copy-editing to a certain extent.
Cliff Guren, with his proclivity for technology, believes in finding out the power-up points where AI can complement and accelerate workflow. He emphasized that this was a wake-up call for all of us to become more conscious of what we do, how we do it and where we think we can add the most value. As fascinated as Stephen S. Power is with the potential of AI, he strongly voiced the learning experience that editorial assistants had while writing reader’s reports which led them to becoming strong editors. His major concern with using AI to get summaries or copy-edits was with the quality of work generated. Even if humans were to check AI’s work, the possibility of errors being ignored by AI remained. Barbara Ruehling seconded Power’s opinion that AI was not great at text production and affirmed that testing was required to find where AI could find its use. At the moment, she felt that it was helping to synthesize the inconsistencies that cropped up when 10-15 authors were drafting a scientific book together and that it would make a great platform to update and localize open educational resources such as college books. They agreed that the question of biases in AI was a nuanced and separate discussion that could impact the publishing industry.
The discussion on how AI and marketing could gel together was spearheaded by Fauzia Burke (President & Founder, FSB Associates), Keith Riegert (CEO, Ulysses Press) and E. J. Wenstrom (Author). This was by far the most optimistic panel which included personnel who were actively experimenting with AI.
- J. Wenstrom talked about using AI for promotion and marketing purposes to get a broad picture overview, ideas for a blogpost, ad copy for Amazon or Facebook, etc. But she also pointed out the need to not use everything generated using AI as it is to avoid content fatigue and turning the world of the written word into an echo chamber. Keith Riegert reflected on how he ensured that his team was well-versed with all things AI by experimenting with it at least an hour every day. Their most prominent use case was to use Adobe Photoshop beta for generative fill of aspect-ratio. but he also emphasized how they were all licensed pictures. Fauzia Burke went over and beyond what she termed as the ‘low hanging fruit’ of AI. She believed AI was a terrific way for her to customize pitch letters for different media houses. AI, when monitored discreetly, could also yield appropriate information by dicing up an author’s bio for different purposes. With the increase in focus on search generative experiences, they all agreed on the earthquake that was about to hit the industry where in they would have to learn to adapt the standard SEO tools to fit the new algorithms of Global Science and Engineering (GSE) that was slowly going to change discoverability on the internet.
Pace University Professor Ken Brooks (President, Treadwell Media Group) along with Bill Kasdorf (Principal, Kasdorf & Associates, LLC) and Diem Bloom (Director of Publishing Operations, John Hopkins University Press) formed the panel that talked about streamlining the journey of producing a book with the help of AI.
On the production side of things, Professor Brooks began with the fact that AI has existed in tandem with various software used within publishing already. Diem Bloom seconded this and explained that on the production side of things, AI could come in handy for generating ancillary content like indexes, and the practice of using software to generate indexes had been going on for some time but it had a large room for improvement. Bill Kasdorf pointed out the major flaw of generative AI as embellishing, notwithstanding the hallucinations which were also a source of concern for tech companies. He said that the right way to use AI is to have human beings in the loop by putting them both at the beginning and the end of the process. The panel strongly felt that instead of taking away jobs, AI would require more personnel to weed out the inconsistencies. Using human judgment was key to using AI, which had the potential to make workflow a lot better. The three aspects that AI could find best use according to them were scalability, for routine mundane tasks, and to get a head start.
Scott Sholder (Entertainment, Media & IP Attorney) tackled the topic of the legal landscape for AI, authoring, and publishing. Being the plaintiff in the authors’ litigation case against Open AI, he spoke from a close familiarity with the major legal concerns of the day. He explained what protectability and infringement signified in legal terms and pointed out how it was not easy to draw lines with the shape of the present law. He also discussed the major aspects governing fair use, which included – the purpose and character of use; the nature of the copyrighted work; the amount and substantiality of the work copied; and the effect of the use on the potential market or the value of the copyrighted work. Sholder pointed out that the first and the last aspects of the fair use laws were the ones that drew the most attention and hence were at the center of all contentions.
The webinar also represented the creatives, the authors whose input would make this discussion complete, as they are the ones facing the challenge of their art being taken over by a machine. Peter Brantley talked to Sean Michaels (internationally bestselling novelist and critic) and Gregg Hurwitz (bestselling author of the Orphan X novels). Both authors emphasized the need for guardrails to keep AI within the realms of control. Sean Michaels aptly pointed out that in an ideal world, where authors would not need to worry about money, AI could be uncoupled from the creative questions as authors and artists could afford to be less insecure. Gregg Hurwitz was more optimistic and placed his faith in human excellence and the need for community. He believed that personalized content would increasingly turn humans into microcosms and that it would not be a viable option eventually. Human beings require a sense of community and belonging to survive. Having worked with AI for his latest novel (Do You Remember Being Born?) Sean Michaels talked about his concerns about AI simulating literature in a way that it was not easy to recognize that it was simulated but he like Gregg Hurwitz was excited by the potential of AI and the question of fake art and what it meant to be fooled by fake art.
The closing keynote speech was given by Ethan Mollick (Associate Professor, The Wharton School) who is an advocate of the practical uses of AI through his blog, One Useful Thing. He termed AI as undetectable, ubiquitous, and transformative. He spoke about the turning point in AI with the essay “Attention is All You Need” which focused large language models more on prediction and the era of predictive creations began. He also talked about the jagged frontier of trying to identify jobs that AI was good at or not. He spoke about how worrying about AI getting smarter than humans while being a valid concern, was a distraction as well as quite an apocalyptic approach. He invited everyone to adapt to AI by following the principles of – inviting AI to everything; being the human-in-the-loop; telling AI who it is and treating it like a person; and that this was the worst AI we would ever use.
In all, the webinar had an in-depth look at AI – how it would be evolving in the future and what it would mean for the publishing industry! The call to be open and adaptable was easily discernible in the entire seminar. Publisher’s Weekly looks forward to taking the discussion further with its next online seminar, AI Tools and Solutions for Book Publishers: Harnessing the Power of Artificial Intelligence, set for February 21, 2024.