Would you read an AI-written novel?

That is the question, more or less, that I posed in subject header of my last email newsletter. The goal was to be a bit provocative, of course, but I did genuinely want to hear from my subscribers on this topic, and in the author’s note I shared a lot more of my thoughts on the subject of AI-assisted writing.

So far, I’ve gotten about half a dozen responses—a lot less than I was expecting, honestly—but the responses I have gotten have been overwhelmingly negative. As in, “no way in HELL will I ever read an AI-written novel, and if you ever outsource your writing to an AI, I will never buy any of your books again!”

…which is awkward, because I’m currently working on an AI-assisted novel, with plans to write several more.

From this and other experiences, it’s become clear to me that there is a small but extremely vocal segment of the population that has strong and vitriolic opinions about AI. There’s probably a much larger but less vocal segment that thinks AI is terrible (though not terrible enough to shout about it from the rooftops), and another large segment that is AI-curious but doesn’t really have a strong opinion one way or another. At this point, the people like my wife who are proponents of AI are practically the lone voices in the wilderness, at least as far as the culture is concerned.

When it comes to books and reading, I suspect that people skew much, much harder to the “AI is so evil!” side of the spectrum than the “AI is so awesome!” side. This is especially true of science fiction and fantasy, since (1) most SF writers are actually luddites in real life, (2) SFWA is a vitriolic echo chamber of the most luddite of them all, and (3) readers of SF&F tend to skew older, tend to be higher educated, and tend to be higher earners than the general population—meaning that they have more to lose with the AI revolution than they have to gain.

But here’s the thing: as an indie author who has been struggling for the last twelve years to build a successful writing career, and has barely been able to keep it going for most of that time, AI-assisted writing represents either a potential game-changer that can help me achieve the kind of success that has eluded me for years, or else it represents an existential threat that will snatch those dreams of a writing career completely out of my reach. There really is no middle ground—at least, not in the long term.

In the old days, there were six major publishers, dozens of reputable small presses, several regional distribution networks, and thousands of local bookstores, where most readers went to buy their books. To have a writing career, you needed to get picked up by a publisher, who would usually give you five or six books to grow into an audience, at which point you were pretty well set up for the future.

Of course, any number of things could happen to torpedo your career, and very few bestselling authors (let alone authors generally) made enough to live comfortably off of their writing alone, but the hardest part of breaking in was breaking out of the slushpile and getting a publishing deal. At that point, you could expect a certain degree of career stability, at least for a few years.

Under this system, it was entirely possible for a genre-specific magazine like Locus to track all of the SF&F books that had been published in the last year. Places like Writer’s Market were able to track all of the publishers, and many readers could—and did—subscribe to all the SF&F short story magazines, and read all the stories. Among writers, there was still a lot of competition, but most of it took place in the slushpile, not after publication.

I’ll be honest: I never actually experienced this system, because it died a couple of decades before my first story was ever published (“Decision LZ1527,” Leading Edge Magazine December 2009). The only things I know about it are what I’ve been able to piece together from Brandon Sanderson’s writing class, Kris Rush and Dean Wesley Smith’s blogs, several now-defunct podcasts like Adventures in Sci-Fi Publishing, Locus Magazine itself, and various conferences and conventions that I’ve attended over the years, including Worldcon and World Fantasy. I’ve definitely done my homework on the subject, though admittedly, it can be difficult sometimes to separate the myth from reality.

From what I can tell, the old publishing system (which was really more of a 50-year aberration, when you look at how things were set up in the pulp era and before—but I digress) began to fall apart when the big box stores like Borders and Barnes & Noble began to take over from all the mom-and-pop indie bookstores. This led to a distributor collapse, since the big box stores only wanted to deal with one or two national distributors. Publishers responded by downsizing their marketing departments, since now they only had to sell to one or two distributors, which in turn led them to drop a lot of authors with small, regional followings in favor of the big name authors with big, national followings. Pretty soon, most of the major publishers were following a blockbuster model, where if your first book didn’t hit big, they dropped you. Then the global financial crisis happened, a bunch of editors got fired and decided to hang out their shingles as literary agents, and pretty soon the only way to get published was to go through an agent first, then make it through the slushpile, then have a national bestselling first novel… and if you couldn’t do that, tough luck.

That was the state of the industry when I first started indie publishing back in 2011. The rise of Amazon kindle and the epublishing revolution gave us an alternative to the soul-crushing, dream-killing system that traditional publishing had become. A lot of us jumped on the chance to “go indie,” becoming our own publishers and digitally publishing our own books.

…except then, Amazon launched Kindle Unlimited and gobbled up most of the market share, making it so you basically had to go through Amazon if you wanted to have a career. Which meant that your career was beholden to the whims of the Amazon algorithm, which favored new books over old books and books in KU over books that weren’t. Also, if anything you wrote fell under the Seattle-based Amazon’s definition of “hate speech,” you could fuggedaboutit. And then, Amazon launched Amazon Ads, which turned indie publishing into a pay-to-play game, where you either had to roll over most of your profits into advertising, or see your books languish with a ranking of 100,000 or higher.

By the end of the 10s, “going indie” was less about being truly independent and more about constantly trying to please the Amazon gods. That meant rapid-releasing, spending a lot on ads, publishing in KU, and writing to trends like reverse harem or werewolf dragon shifter pirates. Those of us who didn’t play that game soon found that we were voices in the wilderness, who were barely able to eke out a meager existence by publishing something new every month.

Now, it is impossible to keep up with everything coming out in your genre. Something like 2-3 million new books are published on Amazon every year, and the bottom third of those never sell a single copy. The competition has moved out of the slushpile and onto the internet, and while lots of great stuff is getting published, the stuff that gets pushed to the top is typically the stuff that aligns with the agendas of the people who control the aglorithms—and in areas outside of the publishing industry, this agenda involves things like drag queen story hour, pride paraphernalia for toddlers, medical assistance in dying, mask mandates and vaccine passports… the list goes on and on.

Which brings us to where we stand today, with generative AI poised to revolutionize the publishing industry yet again. Like it or not, AI is going to change everything—we can already see the wave beginning to swell. The only question is whether we, as authors are going to catch this next wave, or be crushed by it.

It could very well be that this wave is actually a tsunami. That is the pessimistic scenario. If it is, then all of us writers are toast, because the readers of the future will all be amateur prompt engineers who just tell an AI to write what they want to read. A handful of big-name authors will hang on for a generation or two, just on the strength of their brand, and a very small cottage industry will emerge for authentically human-written books, but it will mostly be for hobbyists, like crocheting and perler beads.

If the pessimists are right, then there’s nothing we as writers can do except roll over and die, maybe after vainly shaking our fists at the sky for a little while. That’s what most of the folks over at SFWA are doing right now. But having worked with some of these generative AI tools for several months now, I don’t think the pessimistic scenario is going to play out.

Instead, I think that most readers are going to find that the kind of books they want to read are not the kind of books that they can generate easily themselves. A lot of amateur prompt engineers will have fun with it, just like lots of fans have fun with fanfiction right now. A handful of these prompt engineers will get good enough to generate the kind of books they want to read, and will turn into writers, but that’s not going to be most readers.

Meanwhile, writers will divide into two camps: those who embrace AI-assisted writing, and those who reject it. Except for a few big-name authors who already have a big readership, those who reject AI-assisted writing will find that they cannot write fast enough to keep up with all of the AI generated books and stories that are going to flood the market—not a flood of crappy books, but a flood of passably fair to genuinely great books, as AI technology continues to get better.

Those authors who do embrace AI-assisted writing will find that the AI tools are surprisingly difficult to master, and require a complete retooling of their writing process in order to use them effectively—but after they do, they will find that these AI tools are incredible force multipliers that allow them to write significantly more, and write significantly better. They will be able to rapid release without burning out, and will thus find much more success in building their readership, since publishing a new book is the best way to market all of your old books.

But since (to my knowledge) no one has yet mastered these AI tools, for the next few months/years, most of the AI-assisted stuff that gets published is going to be pretty bad. The whole world is now on a curve, and when we reach the top of it, we will begin to see some really great stuff come out from those authors who are putting in the time and effort right now to truly master these AI tools, and to integrate them into their creative process.

This is why I personally am very excited about AI-assisted writing: because in a world where millions of books are published every year, discoverability is my biggest challenge, and the solution to the discoverability problem ultimately comes down to being more prolific. That is why I try to publish at least one thing every month, usually a free short story… but if I could publish a $2.99 novel every month, that would be so much better. By myself, I don’t write fast enough to do that—but with an AI, I probably could. And it’s not like the discoverability problem is going away—in fact, I expect it to become even more challenging, with tens of millions of books getting published each year as AI-assisted writing becomes mainstream.

Ultimately, though, I think that the key to a successful writing career in a post-AI world is going to involve building a community of fans around your books and your writing. Among other things, fannish communities help to humanize and personalize the connection readers feel with their favorite authors, which is probably why so many readers answered “NO!” to my initial question. The concept of a purely AI-written book probably feels just as threatening to those pessimistic authors as it does to those readers who love that human connection they feel with their favorite authors and fan communities.

But the fundamental reason I’m optimistic about this is because I don’t think there is, or ever will be, a book that is written purely with AI. Even if the author is more of a prompt engineer than a writer, there’s still got to be human involvement somewhere in the process. And if that person is also an experienced writer, who has successfully written several novels of their own, they’re going to be able to leverage that experience in a way that a pure prompt engineer can’t.

So it may turn out that the writers who are best positioned to succeed in the coming years are the ones who cut their teeth in the old world, before the AI revolution, because very few writers in the future are going to have the confidence and experience that comes from writing a novel entirely without AI. Thus, all of those writers who already have a few novels under their belt, and who take the time to truly master these AI tools and integrate them into their process, may be in the best position of all. That’s the optimistic scenario, and that’s the one I’m currently betting on, which is why I’m doing everything I can to master these AI tools.