In the book publishing community, fear and outrage about AI has focused on LLM training on pirated works and how the technology might displace humans. Less attention has been paid to how AI affects search and book discoverability, whether via Amazon, Google, or a chatbot like ChatGPT.
As more people use AI for information gathering—or go no further than the AI-generated summaries at the top of search results—website traffic has declined. Other publishing sectors—news media and magazine media—now prepare for a world where they no longer enjoy organic search traffic. Certain sectors, such as health and travel, have already been hit hard.
How much should book publishers or authors be concerned about AI overtaking conventional search? I talked to three people who’ve been writing about and studying this issue: Ricardo Fayet of Reedsy; Thad McIlroy, an industry consultant who focuses on the future of publishing; and Keith Riegert, CEO of Stable Book Group and Ulysses Press.
Amazon product search is already shaped by AI. Amazon has been testing an AI-powered shopping assistant, Rufus, since late last year. It works by pulling in information from customer reviews, community posts, and other sources deemed reliable by Amazon to answer customer questions. Sponsored ads show up in Rufus too, but they remain in the experimental phase. (Learn more.)
Riegert thinks Rufus is Amazon’s AI-powered-search testing ground but not the end solution itself. First, Amazon is unlikely to revolutionize the current shopping experience, which works well and which customers rate highly. But also an AI-powered model could end up damaging advertising revenue. (Google faces the same problem when generating AI overviews, as people don’t click on the sponsored or organic links as often.)
However, Riegert can envision a future in which, once you upload a digital file to Amazon, AI will “read” or ingest the book and use that to improve discoverability and search. “[The AI] determines whether it’s a good book or bad book,” he said. The upside: It could dissuade people from uploading AI slop to the marketplace if such titles are never surfaced in search. But the downsides are obvious. “You’re allowing AI to essentially determine whether books are worth buying,” Riegert said.
In the spring, authors expressed anger at Amazon’s generation of brief recaps for series to help readers recall what happened in previous installments. Authors weren’t asked for permission, nor can they opt out of the feature. But if Amazon takes it a step further and ingests work for the purposes of better search and discoverability, authors might be foolish to opt out, even if it were possible. Riegert said it could end up being like Amazon Advertising: “If you don’t want to do Amazon Advertising, that’s fine, but it’s going to make your life a lot harder. Amazon is very good at turning things that were optional into things that are mandatory.”
To be clear: There are no signs Amazon will soon be ingesting entire works to improve search results, but it’s easy to imagine, since so many of its features are AI powered. In an article for the July/August 2025 issue of IBPA’s magazine, Riegert discussed the latest best practices for optimizing book product pages on Amazon and emphasized that the latest Amazon algorithm (known as A10) is better at reading product page descriptions and assessing how readers feel about your book. Rather than focusing attention on off-page keywords that remain unseen by customers, authors and publishers should focus on readable descriptions (especially the first 100 words), quality A+ content (with alt text for visuals), and visible metadata—information about a book that appears on the product page.
McIlroy worries that publishers remain behind on optimizing their book’s product pages on Amazon as AI now threatens to move the goalposts. “The way publishers currently approach metadata broadly and SEO is hopelessly inadequate. In their minds, they’ve done a pretty good job. If you filled in all the little boxes on Amazon, you’ve done metadata, with this fixation on keywords that’s been there forever. It’s like ‘Get the keywords right,’ that they are some kind of magic. They’re not.”
Ultimately, because of AI, McIlroy believes product metadata on Amazon will stop mattering entirely. “It’s there, it’s a little marker, but all that matters is the content, all of which can be subsumed in a way that the machine understands it.”
What about AI-powered search outside of Amazon? AI-powered recommendations have wormed their way into all corners of the internet, and AI overviews already appear at the top of Google search results. While declining website traffic due to AI overviews or chatbot use isn’t seen as an existential threat for books like it is for online media, it will change book discoverability.
AI chatbots such as ChatGPT can make specific, accurate recommendations based on analyzing dozens of different sites. Fayet at Reedsy recently discussed his results from a deep research prompt asking Google Gemini for romantasy recommendations: Here is the full output, complete with citations. Among the sources are Amazon, Goodreads, Reddit, NetGalley, and BookBub. You’ll also find Penguin Random House and a wide range of bookstores and blogs cited.
People who specialize in SEO (search engine optimization) now seek methods to influence how brands or products get mentioned in such responses from AI chatbots—an emerging field known as GEO (generative engine optimization). McIlroy says, “That’s a big, big, tricky issue. ChatGPT’s search right now does actually use a Google-style index of the web. If your amazing SEO has you on the top page of Google, that will probably come into AI search … but that query has to be relatively specific for you to show up.”
Some startups are investing millions on their bet that chatbots can be influenced to mention specific brands or products, while others think it is folly. You can already find obvious attempts to manipulate LLMs in this search, which shows how people have instructed LLMs through invisible text not to be negative, only positive. But Fayet believes GEO will prove much harder than SEO. “Right now, with traditional SEO, it’s almost been too easy to track, because Google gives us a lot of data, tells us exactly how many hits we got on which pages, which keyword searches they came from, what our average position is for those keywords, what our click-through rate is for those keywords, how many impressions we got. It’s a lot of data, and we’ve become used to it.” But with AI, it will be challenging to measure results, because people prompt or query AI in ways that can’t be boiled down to keyword phrases. “Search is going to become fragmented in terms of the queries we use, and also the results are going to depend on LLMs that are trained on our personal data. So the answers are already very dependent on who asks the question,” Fayet said. It’s easy to see how this ties right back into AI-powered Amazon search, where its analysis of your reading, review, and shopping history affects search results or recommendations.
Bottom line: For book discoverability, keywords appear destined to drop off the map in terms of impact or importance, but too much remains unknown about the future of AI-powered search to make specific recommendations for authors and publishers hoping to optimize. Fayet said, “What I think would be really beneficial is to start playing with LLM search. Run the kind of searches that readers of your books would likely run right now or in the future.”
Riegert said, “I think the bigger threat is, especially for publishers like us at Ulysses Press, where we specialize in nonfiction, a lot of our books are topic-specific and meant for people who want to learn a lot about a topic. ChatGPT can probably teach you everything in those books. … I don’t think [such use] has really ramped up that much yet, because there’s still a huge reader base that is not actively using ChatGPT and not necessarily trusting the results that are there. But it is going to make nonfiction publishing a lot more challenging.”
Further reading
- Large Language Models Are Not Search Engines by Dave Friedman
- How Generative Engine Optimization (GEO) Rewrites the Rules of Search by Zach Cohen and Seema Amble (Andreessen Horowitz)
- Hitting the Red Button by Brian Morrissey, about declining website traffic for web-based media and potential remedies in the form of pay-to-crawl schemes
- The Precipitous Decline of Brands by Brad Berens (scroll past the opening link roundup)
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.
