Welcome to the age of algorithmic fiction. Thanks to tools like GPT-4, a human writer with a library of previously written books can simply write a one-page outline for a new novel, and AI can write an entire novel in their style. In many cases the book will be good enough to pass as if it was written by the human authors themselves, allowing writers to publish more frequently. However, GPT fiction will only work with formulaic writers whose books are all similar. In other words, those writers who were already writing algorithmic fiction before the aid of AI.

Commercial fiction writers are “commercial” because their books follow a familiar formula that has been successful in the past. You know the type—those who, when you walk into a bookstore or library, there are entire shelves devoted to that one author, all with covers that have the same style typeface and font, just with different titles. These authors work in popular genres like romance, crime thriller, cozy mystery, or urban fantasy. Their readers expect a certain recipe to be followed—it’s just the specific ingredients that change in each book. Swap in a new McGuffin, love interest, client, crime, antagonist, and setting. The protagonist will face some setbacks along the way but in the end solves the mystery and/or falls in love, often with a teaser to set up the next book. These types of formulaic stories are the easiest fiction to automate with AI. The writer only needs to update the ingredients in a quick one-page outline, then a GPT could write the rest of the book for them in the style of the previous books in the series. 

Many commercial authors who write in ongoing series are already prolific. They publish 1-2 new books per year. Or if you’re like James Patterson and use co-authors, they might publish 5-10 new books per year. Commercial self-published authors are also prolific, some publishing a new book each month. AI can essentially serve as a co-author that you don’t need to pay, allowing the high-frequency self-publishers to publish even more frequently. AI text-generators could allow such authors to publish a new novel weekly—or even daily.

Personally, I wouldn’t be interested in reading such AI-generated books, just as I’m not interested in reading the formulaic commercial fiction that was written by human authors before AI. What I enjoy reading are unique literary works that do not follow a familiar formula, where you do not know what will happen next. I like books that are genre-defying, that make you say, “I’ve never read anything like that before.” Or perhaps there are other things like it, but not so much so that the book could have been predicted by my brain—or by an AI’s “brain” (aka Generative Pre-trained Transformer).

Human writers who want to compete with AI must aspire to write unfamiliar fiction, or novel novels. Thankfully that is already the kind of fiction I like to write. To quote myself: “If you can be replaced by ChatGPT you should be. It means you’re not writing from a unique perspective. You’re already a bot.” Stop following a formula and start following your muse. Your co-author should be your subconscious, not ChatGPT—or a human ghostwriter. 

I don’t blame commercial fiction writers for doing what they need to make a living; it is their readers I don’t understand. Who are these people who read multiple novels every day, constantly craving more formulaic plots featuring the same characters, over and over again? I guess they’re the same people who keep Law & Order, CSI, NCIS, and their myriad spinoffs perpetually atop the Nielson ratings charts. They get turned off when they see or read something “too weird.” There is no such thing in my book—the weirder, the better. If it’s not weird in some way, I am not interested. 

The real issue for writers like me is finding enough readers who desire unformulaic fiction—weirdness and novelty, true creativity. Judging by the bestseller lists, that was rare before AI writers. It will only become more difficult for non-algorithmic fiction to stand out as the market is flooded with billions of mediocre GPT-written books.