What is the future of publishing in a world of artificial intelligence? (What’s the future of anything these days?) Will Anderson of Penguin Random House in New York says he sees two things: better written non-fiction books and novels more centred on feelings. In other words, non-fiction books written with the assistance of AI models published alongside the type of novels and memoirs that AI can’t write: books imbued with memory and physicality and all the emotions people feel that AI can’t.
At least so far.
Anderson is a senior editor specializing in Christian books at Penguin, and he was speaking at a recent online seminar organized by Substack strategist and writer Sarah Fay. The seminar was centred on marketing—how writers can sell new books in today’s rapidly-changing market, and how we can use evolving opportunities to re-release older titles. At the end, Sarah asked panelists to talk about the future of publishing, and Anderson brought up AI.
I always think of “predictions” as descriptions of the present day made by experts who can see things already underway that most of us haven’t yet perceived. People who actually try to gaze into the future often seem to think trends will continue in a straight line, and since they don’t, futurists are often wrong (Warren Buffett aside).
Yet Anderson talked about something he says is already starting to happen with non-fiction books, and it’s a trend he believes will burgeon. Astronauts, doctors—movie stars for that matter—are experts in their fields, but they’re not necessarily good writers. In the past, celebrities often worked with ghost writers. I had fun a little while ago writing about Prince Harry’s collaboration with a ghost writer on his memoir Spare, a man whom Harry only scantily acknowledges.
Now, says Anderson, publishers can work with their experts and celebrities on detailed, chapter-by-chapter outlines. When they’re satisfied, they can feed an outline into an AI model, and the machine can churn out a manuscript at speed. (Bye-bye ghostwriters.) Afterwards, of course, AI-generated manuscripts have to be closely edited. But Anderson says the quality can be high, and predicts that AI will raise the bar for future non-fiction titles.
“The average writer will sound better than they did before.”
Anderson’s thoughts about novels and memoirs are more speculative. But he said he expects a demand for books that are more feeling-centred, opinionated and personal—“something AI can’t regurgitate.”
Essentially, he’s talking about human authenticity, and that’s something I’ve been thinking about since attending a recent talk by novelist Samantha Harvey. Harvey was interviewed onstage at the Toronto Reference Library by journalist Rachel Giese about her novel Orbital, a quite wonderful international bestseller that won the Booker Prize in 2024. It’s a book that couldn’t have been written by AI, and her talk made it clear why.
That night in March, Harvey came across as gentle and humorous. Someone I know met her when she taught for a semester at the Humber School of Writers a few years ago, and said she was uniformly kind to students. Sitting onstage, facing a packed auditorium, Harvey was also open about her creative process. I think it’s safe to say there’s nothing machine-like about it, even though Orbital is set on the International Space Station, high in orbit over our threatened earth. (I reviewed the book here.)
In fact, Harvey seems to have brought a combination of intuition, hard-headedness, writing chops and a bad case of insomnia to writing the novel, which makes for a very human approach. Of course, every writer has a different process. I’ve always felt terrible for students who arrive in class with puppy-dog eyes and a copy of Stephen King’s how-to guide, On Writing. King says writers should be able to draft a novel in three months, meaning that he can. Most people can’t, and they won’t be able to do what Harvey does, either. The trick is to find what works for you.
Harvey said that when she was first thinking about writing a novel set on the space station, she checked around and was relieved to find that no one else had written one. “I have a terror of writing a novel then finding that someone else has already done it.” After she felt able to start work, she said she had to think about how she was going to “alchemize fact into fiction. I didn’t want the book to feel too essayistic. It was a novel, after all. So the question was: How do I create a spell that won’t break until the novel is over?”
What did she have to avoid?
“Sometimes fact, especially technical fact, can break the spell.”
Giese was curious about Harvey’s research. She pointed out that Harvey had clearly looked into the minutia of daily life aboard the space station for the four astronauts and two Russian cosmonauts cocooned up there, each of whom stays in orbit for about six months. Did she discover anything astounding that she hadn’t thought about before?
Harvey laughed. “Everything. You have to have training courses to learn how to use the toilet. Two courses. The Russians until the last couple of years didn’t have their own laboratory on board. And because they didn’t have one, they spent their days playing chess, floating around and growing lettuce.
“They all do a lot of vacuuming, cleaning, and finding things in the filters. It’s very domestic and very mundane, doing such quotidian things to a very set schedule while orbiting the earth.”
It occurred to me later than an AI model couldn’t not know any of the eccentric facts that delighted Harvey as she unearthed them. It would know all about everything relevant before it churned out a manuscript, which presumably means that Harvey’s sense of surprise and discovery aren’t available to an AI model; that a machine can’t demonstrate her very human delight in the new. How can anything seem new to an AI model that has access to all human knowledge and culture? Or will it eventually learn to believably mimic our feelings as well?
It was a night of questions—mine, Giese’s, and others from the audience. Geise was the one to ask Harvey about her research on the astronauts themselves.
“You can never really get to what astronauts think about things because they’re not going to tell you,” Harvey replied. She said she saw the International Space Station as a model of international cooperation—“a beautiful symbol of peace”—and one that was disappearing as she wrote about it, given plans to shut it down.
“But I was speaking to a NASA astronaut about this. I said I felt sad, and asked if she did. She said, No, not at all. Astronauts are almost pathologically optimistic. She could see doing bigger things. Going to the moon, to Mars. When I think, why bother?”
Another laugh: “There was a lot in this book I had to make up.”
As to her writing itself, Harvey told Giese that it was heavily influenced by her serious case of insomnia, which she wrote about in her memoir, The Shapeless Unease: A Year of Not Sleeping, published in 2020.

“There’s so much in common between Orbital and the insomnia book,” Harvey said. “I’m sure, for one thing, that insomnia helped me know what it’s like when day and night are upended.
“I’m sure it’s not good for my brain, but it’s quite creatively useful. It’s given me a manic focus. Since having insomnia, I write in a more concentrated and focused style, It’s shorter.
“In a way, writing is my way of dreaming.”
The act of writing is also so deeply central to her sense of self, she said, that she insisted on writing while suffering badly from sleep deprivation. “I wanted to harness that instinctive energy. That’s perhaps why (Orbital) has a greater variety of voices and tones than my other novels. I wanted to have the freedom to go where I wanted to go—zoom in, zoom out of different characters, to try to expand and contract time. And I think all of this came from the new wiring in my brain that came from lack of sleep.
“The book for me is an emotional project rather than a cerebral project,” she said. “It’s born of feeling—love for the earth and the natural world. A feeling of awe, of being overwhelmed by the earth. You see this tiny thin layer of atmosphere that makes life possible. (Writing Orbital) was an expanded, oxygenated feeling of being in love. I don’t know if it’s a spiritual project, but I am, or at least I’m trying to be alive to experience, and I wanted the book to express that.”
Meanwhile, she composed it, as she had to, in the way that works for her.
“The actual process of putting sentences and paragraphs together is for me the really joyful part. In rewriting, it’s always the bigger things I have to put together. The structure level. The pacing level.” Laughing about her first drafts: “I tend to end up with 80,000 words of fine sentences in a structure that makes no sense at all.”
A decade ago, publishers were beginning to worry about the proliferation of self-published genre novels appearing on Amazon. These were leaching away the sales of the romance novels that elevated the bottom line of many publishing houses. Given the ease of Amazon’s new Kindle technology, self-published authors could offer their books online for $4.99 and less, having no editorial or printing costs, and publishers were unable to compete.
At the time, I was curious and downloaded a couple, but managed to read only a few chapters of each, they were so badly written. This didn’t seem to bother readers; I’d chosen bestsellers. But a combination of the low prices, ready availability and the sorry state of literacy, especially in the U.S., meant that people didn’t seem to care.
Now, AI-authored romance and sci fi novels are another new trend online. Some self-published writers are crafting outlines, then getting bots to perform the slog of actually writing their books. Other AI-authored titles are being released by non-writers wanting to exploit the self-publishing market. The books don’t have to be well-written. In fact, they probably can’t be. A decade of bad self-published books has conditioned readers to expect very simply-plotted, predictable and repetitive stories that help them get to sleep at night, and AI is able to do that already. Will Anderson of Penguin didn’t mention them, but self-published AI books are another challenge to the future of both writers and publishing—although it’s true there have always been bad books published, and we’ve survived them before.
I went to Harvey’s talk simply out of interest, but I took Sarah Fay’s seminar because I’ve finished a good draft of the novel I’ve been working on for a couple of years. This means it’s time to research the changing marketplace in which I’m going to try to sell it.
I find it best not to think about the market while writing. You can end up chasing trends that will be passé by the time you finish your book. I’ll have to rewrite the manuscript again (and again, and again), and do additional research as the need arises. But after spending a couple of years lost in the pleasures of writing, it’s time to throw another type of research into the mix.
I can only hope Anderson is right, and there will still be room for literary novels.
Won’t it be interesting to find out?