AI-Generated ‘Summer Reading List’ Is Bad News for Readers and Writers

Sun-Times' laughable mistake caused real harm for book authors and readers.

Photo by Annie Spratt on Unsplash

It’s no secret that AI text-generators have stolen opportunities from human writers. Children’s books, TV and movie scripts, news reports, short stories, poems, social media posts—the list of written works that AIs can generate goes on and on.

Readers may dislike AI content, but it’s not clear that publishers care. For them, text that’s faster and cheaper to produce, even if it’s rehashed, regenerated, and littered with factual errors, may be just as fit for their profit-making purposes as stories written by humans.

Recent evidence of publishers’ cost-cutting mentality and general disregard for human writers comes from the Chicago Sun-Times’ May 18 publication of a “Summer Reading List,” which featured 15 books, 10 of which were AI hallucinations. Yes, you read that right: 10 of the 15 recommended books didn’t exist. For the record, Percival Everett never wrote a book called The Rainmakers about a futuristic drought in the American West. Nor did Isabel Allende write a climate fiction novel called Tidewater Dreams. Everett did write James, Dr. No, The Trees, Telephone, So Much Blue, Erasure, and I Am Not Sidney Poitier, and Allende did write The House of the Spirits, Paula, Maya’s Notebook, A Long Petal of the Sea, The Soul of a Woman, Violetta, and other works.

You may have heard about Sun-Times’ reading list fiasco, first reported by 404 Media. The Sun-Times itself later explained that the reading list was part of a 64-page supplement produced by King Features, a news content syndicate that the Sun-Times repeatedly characterized as a “third party.” That characterization is true in the sense that the Sun-Times doesn’t own King Features. It’s false in its implication that the Sun-Times wasn’t responsible for the reading list or the other errors in the supplement it mailed to its subscribers.

Rather than take responsibility for foisting fake content on its readers, the Sun-Times blamed King Features and a King Features’ freelance writer, who the newspaper described as the “author” of the supplement. This “author” was a real person, but once the errors were discovered, he admitted that he didn’t write the reading list. He used AI to generate it and somehow failed to notice that the books didn’t exist. (Show of hands, who thinks he’s now using AI to “write” a memoir about his mistake that went viral?)

An apology written (maybe) by Chicago Public Media CEO Melissa Bell is worth reading not because she promises readers that the Sun-Times won’t use AI in the future, but because she blames “human mistakes,” in other words, real people, primarily King Features and the freelance writer, for the errors in the supplement. She also blames the newspaper’s circulation department, which managed the supplement as what she called a “creative solution to keep hitting our revenue goals.” For Bell, these “human mistakes” weren’t a total failure of editorial integrity and oversight; they were necessary and justifiable attempts to boost the company’s bottom line.

“It’s easy to say AI is a problem,” Bell wrote. But then she didn’t actually say that AI was the problem. Instead, she said, “It’s a lot harder to work, collectively and individually, as humans to catch up and learn and understand how our industry and technology are changing around us.” In other words, the fake content wasn’t the AI’s fault, it was the people’s fault because they didn’t understand how the world now works. Let’s be clear: people did make mistakes, but without AI’s involvement, those mistakes couldn’t have happened.

What does all of this means for writers? The answer isn’t clear, but I’ll hazard four predictions:

1. AI companies will win. Publishers will continue to scale up their use of AI-generated text not because it’s better or more accurate than writing by humans, but because it’s cheaper and faster, their shareholders expect them to use it regardless of the problems it creates, and they no longer trust human writers, especially freelancers, to produce AI-free work.

2. Writers will lose. Many writers, and again, especially freelancers will be forced to use AI because they won’t be able to compete against it. It’s just too fast, too cheap, and too tempting. Writers will use it for brainstorming, researching, drafting, preparing summaries and synopses, revising, editing, and a myriad of other writing-related tasks. (Two years ago, I used ChatGPT to generate a title for an unpublished novel with hilarious results. Today, I don’t find that incident nearly as funny as it seemed then.) Writers will tell themselves AI is “just a tool” and it’s okay to use it as long as they’re careful. Then one day they’ll be tired or distracted, they’ll get sloppy, some AI-generated text will escape the cage they built for it, and to their dismay and regret, it will end up in their work unless they decide now to never use it again for any reason whatsoever.

3. Readers will be cheated. Instead of original articles written by real people, more and more of what’s available for them to read will be AI-generated or AI-translated at least in part. It’s already difficult or impossible to tell whether some texts were generated by an AI or written by a human. Readers have already found AI prompts in published novels. Finding factual information that’s free from hallucinations online is already becoming increasingly difficult.

4. Authors will be cheated. The fake “Summer Reading List” was amusing up to a point, but it also caused real harm. Authors were said to have written books they never wrote. Readers were encouraged to buy books that didn’t exist. Real books that might have been recommended weren’t. Authors of those books missed out on readers they might’ve had. Authors and their publishers missed out on book sales. Opportunities for real conversations among real readers and authors about real books were lost.

Over time, AI will make more mistakes, not fewer. Soon, these mistakes won’t be newsworthy. No one will apologize or bother to correct them. Readers will be expected to assume mistakes were made, but no worries – it’s just the way things are and it’s good for the economy, right?

That’s a bleak assessment, but I still hope that there will be a backlash and that writers, readers, and publishers will say no to AI. Right now, every refusal to use it is a victory for human writers—and humanity. #