In the second part of our coverage on PW’s AI webinar, Artificial Intelligence: Revolution and Opportunity in Trade Publishing, held September 27, we look at first-hand accounts of how the technology is impacting three core publishing operations: editorial, marketing, and production.

In the Weeds with Words

The editorial landscape continues to be in flux, and with the introduction of generative AI, many professionals are faced with the dilemma of ignoring or embracing new technology. Three panelists—Stephen S. Power, executive editor of Kevin Anderson & Associates; Cliff Guren, founder and CEO of Syntopical; and Barbara Ruehling, CEO of Book Springs Ltd.—took part in a conversation that shows that various forms of AI are already being used by editorial departments.

AI has already been used and tested as a means of ironing out the editorial workflow, panelists said, be it finding ways to save time or how AI could be used to organize data in a more efficient way. Still, Power pointed out generative AI’s major flaw: “What I’ve found is that it’s just not good at writing stuff. It’s very good at coming up with mediocre language because that is what it is literally designed to do, to come up with most generic possible answer to a question.”

One fear Power shared is that if editorial assistants use AI summaries to help them create marketing copy or write information for the tip sheet rather than read the entire book themselves, they will miss out on nuances in the book. Without making editorial assistant develop the necessary skills to summarize plots, publishers run the risk of preventing assistants from learning how to become editors, killing of the editorial minor leagues, Power said. He is also wary of publishers using AI to filter book submissions, afraid that relying on algorithms will kick out worthwhile manuscripts that “don’t fit it,” simply because they didn’t have some key words or phrases.

Ruehling agreed with Power that available AI tools are “not that good at text production.” She talked about how her company, which works with teams of authors, has begun testing AI, finding it helps with the more repetitive tasks, providing more time for the writers to work on creative tasks. “If we turn our backs to AI, the writers in our group will still use AI,” and likely produce inferior results, Ruehling said.

Guren predicted that publishing houses will eventually develop their own large language models (LLM) based on their author lists that they will use for marketing purposes. “AI is a wakeup 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,” explained Guren.

Where AI and Marketing Meet

In marketing, marketers are quickly learning how to use AI as an asset rather than see it as a worry. In the session, panelists Fauzia Burke, president and founder of FSB Associates; Keith Riegert, CEO of Ulysses Press; and E.J. Wenstrom, a fantasy and science fiction author, offered their individual outlooks on book marketing now and in the near AI-affected future.

“I spent years learning how to code in Python,” said Riegert but when he asked ChatGPT to scrape a retailer, “it wrote better code faster than I could even type it and I realized I had wasted two years of my life,” Riegert joked. Wenstrom knew things were different when she started having “full conversations with ChatGPT and saw what the AI voice and conglomerated insight might look like.” Burke said though she has been working in book marketing for over 20 years, “when ChatGPT was announced, I became an early adopter.”

The marketing professionals were enthusiastic about sharing the tools and experiences they’ve had. “Jaspar AI and Writer AI are getting a lot of engagement right now,” said Wenstrom. Among a range of possibilities, both AI tools can provide everything from ideas for a blog post, marketing plans and pitches, press releases, even copy that can be used for Amazon ads or social media posts.

Burke spoke about how AI can help with parsing and exploring internal data sets. “I’m sitting on a data set where we’ve kept track of online media for the 20 years,” she said. Where she had previously had to search and analyze the data manually, Burke is now using AI to build a program that can use the data set to create customized press pitches. Other tools like Otter.ai, which creates text transcripts from audio and video, and StudioShot, a tool that takes selfies and turns them into a studio quality picture, were mentioned as being useful for marketing materials.

Even with the shared excitement, the panelists agreed it is wise to proceed with caution. “These tools are going to change so quickly,” warned Riegert. There’s the sense that in the next three months to a year, the tools in use will function entirely differently.

Riegert pointed out about a big change to how books will be discovered and marketed: “We have to touch upon one big earthquake that is coming our way and that is search generative experience (SGE).” Search generative experience utilizes AI to give applied search results. Where searching Google right now produces a list of sites, you can glimpse SGE in action on Bing. Riegert offered an example of how searching using Bing’s AI for, say, “the top 5 summer reads of 2023” it will give you five results whereas a typical search would give a much longer list. Google is testing its own system, which could reduce organic search on website by 25% to 50%, Riegert said. “What I’m really nervous about is that Amazon is currently working on their own SGE system,” Riegert said. “As marketers we’re going to have to replace our standard SEO tools to adapt for the new environment.”

Burke suggested it is best to not leave AI in the hands of IT people, noting that was the case with the internet and social media, and it took a while to integrate everything into the creative side. “With AI we don’t have a lot of time,” Burke said.

Streamlining the Production Journey

Automation is a key element of the ever-evolving book production toolkit. Three panelists at the frontlines of book production and generative AI discussed the present and future. “We’ve kind of already been doing it for awhile so it wasn’t a big thing than, say, early in my career when e-books or the internet came around,” said Diem Bloom, director of publishing operations at Johns Hopkins University Press. The other panelists echoed Bloom’s sentiment; aspects of artificial intelligence, much like task-based production, has been integrated in production for years. “I’ve used AI and machine learning for years,” said Ken Brooks, president of Treadwell Media Group.

From machine learning to deeper learning, generative AI is being incorporated into practices as another way to help with efficiency and scalability. “It’s being incorporated into tools that you already use,” explained Bill Kasdorf, principal at Kasdorf & Associates, LLC. “It’s probably already in Word or Photoshop and is commonly used in production.” Beyond that, Kasdorf said publishers need to know that their production vendors are preparing to incorporate AI into the larger production process. In connection with that, Kasdorf said, “standards are overarching everything we do.” He added that since standards change all the time, AI can help align those changes.

Publishing production departments are actively assessing and doing tests, especially when it comes to discoverability. “It’s not so much about creating the content but rather cleaning it up,” said Bloom. “We work with so much data and content that when it comes to production, the clock is always ticking and decisions need to be made.” Whether it’s creating an index, metadata, keywords, or even alt-text, panelists agreed that generative AI has only helped with mitigating the task, especially when it comes down to production deadlines.

Brooks predicted “time-to-market will shrink considerably, and the quality of output will increase dramatically; the cost of production come down significantly as well.” He agreed with Kasdorf that AI will be subsumed into the production toolset: “All the tools we use, AI is going to disappear into those tools and they’re just going to become more and more productive.”

Kasdorf outlined three ways production may benefit from using more AI: scalability, doing routine, mundane tasks, and for getting a head start on a task. Kasdorf observed that he had not heard the term “human-in-the loop” used much in the webinar up to his session, noting it is a common term used in the field. “The right way to use AI is A, have a human-in-the loop, but have a human at the beginning of the process and at the end,” said Kasdorf.

Bloom also said she doesn’t see AI eliminating jobs. “I don’t see it as we’re all out of a job,” Bloom said. “Really it’s more like we’ll have more work to do because of this.”