Exploring AI Part 1: changes in publishing

Content alert: the following was written by a human.

In this three part series exploring AI, Books Forward is chatting with Dr. Andrew Burt, author of lots of published science fiction, including his newest novel, “Termination of Species,” for those who like AI, biotech, chess and a bit of romance.

Dr. Burt was vice president of the Science Fiction and Fantasy Writers Association for several years. He heads Critters, the first writers workshop on the web and home to other writerly resources. He runs ReAnimus Press and Hugo-winning Advent Publishers, helping award-winning and bestselling authors breathe life into great books. Outside of writing, he’s been a computer science professor (AI, networking, security, privacy and free-speech/social issues); founder of Nyx.net, the world’s first Internet Service Provider; and a technology consultant/author/speaker. For a hobby, he constructs solutions to the world’s problems. (He jokes: Fortunately, nobody listens.)

PART 1

When ebooks emerged, experts predicted “the end” of traditional book publishing. Is text generative AI “the end” of traditional book writing and promotion?

I was a fan of ebooks since the late 90s, when the very first ebook devices were coming out, and I largely only read only ebooks. As such—plus being a computer science professor and science fiction writer, inevitably peering into the future—I was indeed one of those predicting significant changes to how people read, and thus, to publishing.

However, my predictions came with a caveat, one which has not (yet) materialized: I said that when ebooks could mimic the most relevant aspects of a paper book, they would likely replace paper books. The tipping point I proposed was that when there a device that looked like a book, i.e. several hundred sheets of pages with words and images on them, bound at a spine, the “codex” format, but digital—basically a bound set of hundreds of sheets of thin digital paper, that can each act as a screen displaying whatever pixels we want—and super cheap—then there would be no real need for print books. There is something that people inherently like about these objects we call books: the shape, the multiple pages that you can riffle, that you can stick a finger in to hold a place, that you can quickly flip forward and back in, that can be dropped in a bathtub without causing (too much) emotional damage, etc. (and, as is almost invariably pointed out, “the smell”). But make the pages digital instead of static text, and wow, that would be a game changer.

Ebook devices don’t mimic that today. They display one page at a time and navigation within is cumbersome. The bound, multiple page aspect of a book is critical to its success. This “codex” format pretty much replaced “scrolls” of old. It’s a better format for humans to use. So, the market for print books still thrives.

On the publishing side, it’s not ebooks per se that have altered publishing, but the inexpensive cost of anyone being able to publish a book—both as an ebook and as a print-on-demand (POD) physical book. The one competitive advantage that traditional publishers still have that small or self publishers don’t, is the massive marketing dollars. This creates not just the ability to run ads and do all kinds of promotion to lots of eyeballs, but also the ability to print thousands or millions of copies in advance and get them on store shelves across the country and world. Thus, new authors with a traditional publisher still don’t get access to those huge advantages, and this is where publishing has changed. Now if you submit to the major publishers and get rejected, you now publish it yourself for a lower cost. It may not sell many copies, but it might. Lightning strikes. And the sheer volume of such books does, in aggregate, take readers away from books from the major publishers. A small number of readers each of millions of self-published books represents millions of readers who aren’t reading books from the majors. Readers have vastly more choice.

So—and I’m really working my way back to AI!—the new technology of ebooks, POD and free platforms like Amazon’s Kindle Direct Publishing (KDP) did change publishing, but only when they meet the needs of the consumers. Likewise with AI.

So how do you anticipate AI will affect the industry exactly?

AI is really taking off recently in terms of capabilities, mostly because the computing power has reached the point it allows the massive amounts of calculations necessary.

It’s worth briefly mentioning how these “Generative” AIs work (also called “Large Language Models”, or LLMs; because they were trained on massive amounts of text). They look at a huge amount of text (and images, etc.), mush it up into hundreds of billions of numbers, and spit out the most likely next words based on the clues it finds in a question you ask it. It’s all based on the probability of what word could come next in an answer.

What we see today is the tip of the iceberg in what these AI systems can do. But, looking at what they can already do today, we can (at last!) get to some kind of answer to the question:

When AI can do what consumers-of-writing want, and do it better than humans can do it (notably in the realm of cheaper and faster, of acceptable quality writing), then, yes, AI will be “the end” of that kind of traditional writing.

But note I put some weasel words in there: “that kind” of writing.

There are all kinds of written material out there. Science fiction novels, fantasy short stories, scripts for rom-com movies, self-help web pages, historical scholarly journal articles about the crusades, scientific papers describing experimental results in quantum physics, holiday greeting cards, ad copy for TV commercials about cat food, headline news articles, business proposals, answering support questions in a chat session, getting ideas where to visit for a road trip, job descriptions, condolence emails, instructions for assembling a bookcase… the variety of “writing” is just massive.

Generative AIs like ChatGPT, Gemini, etc., can do some of these things already to the point where paying humans isn’t needed. This of course applies beyond words, to any kind of content, images, videos, audio, etc. These are human “creative” endeavors, even if Ikea pictorial assembly instructions aren’t exactly heartwarming art. 🙂

And that’s a big divide: Content created for art’s sake vs. content for commercial purposes.

I can say with near certainty that AIs will be “the end” of many of these human content-creating endeavors as career type jobs. Chatbots are already replacing human jobs for tech support. (Maybe not well, but they’re so much cheaper that businesses can’t help themselves.) AI created artwork can already replace costly human artists. There won’t be any aspect of “writing” (or other content creation) that won’t be touched by AI in some way. It’s simple economics: When a free/cheap method of producing something is discovered that mostly gets the same job done, it mostly replaces the old more expensive method. Cars replaced horses. How often do people mail hand-written letters to each other? Or send faxes? etc.

If/when this will happen to writing (and promotion) will depend on the kind of writing and promotion we’re talking about. But there are so many kinds of content creation that people are paid for now that simply will be cheaper to replace with AI generated content.

In terms of book writing specifically, how does AI come into play?

At book length, it will take longer. AI can’t quite today write a horror novel better than Stephen King—most AI generated novels are pretty horrible—today—so Stephen King’s job is safe for a few years. Maybe 5-10? If we’re talking non-fiction, since generative AI’s are untrustworthy for factual correctness, they could maybe produce readable prose, but you’d have to carefully check every single fact. AI might be able to scrounge up some actual facts (that you verify) that you didn’t know about, since they’ve been trained on massive amounts of data, although there are a lot of older books and articles that aren’t scanned yet, thus not available to AIs and only to humans. So, I don’t foresee AI replacing book length non-fiction for some time (although it may help make it much easier/faster to write).

At short length, AIs can already create poetry that humans can read and think has deep meaning, even when the AI creating it had no such intention. For that matter, I wrote a really simple AI back in the 90s that created a particular kind of gibberish writing derived from some input text. [It’s at critters.org/bonsai.] It was only a few lines of computer code, but it produced some random, intriguing sounding output that was so odd it often seemed “deep”—simply because humans are good at finding meaning in randomness. Think of how we find animal shapes in clouds.

When we add in the concept of collaboration, where a human author uses AI as a tool to help them write a “better” book than they could have alone, that opens up even more options for AI to replace human effort. Just like spreadsheets made it possible for non-techies to do some complex tasks, people who are less capable at writing can use AI to create books or other length content, especially (non-factual) non-fiction, like a book of jokes (to the extent facts don’t play a role). An AI is suited to taking stilted, grammatically incorrect, typo-ridden prose and making it sound much better. There’s a large market for that. But this could be a double edged sword: In this case, more newbie human “authors” benefit from being able to write “better”; but flooding the market with magnitudes more competent texts means even fewer sales for authors who have the native talent for writing and don’t need AI help. This isn’t to say anyone can ask ChatGPT to “write a 300 page book on the evolution of dinosaurs” and presto! out pops a bestseller, but AI might cut down their effort by a factor of, say, ten. (And who knows, in 5-10 years…)

People write a lot of short- and mid-length reports on things, like news articles, progress reports, job descriptions, etc., where beautiful prose isn’t a goal. If factually verified, AI’s can probably take over a lot of this work. Then the human becomes mostly the creator of the prompts, provider of the data to describe (both of which can be somewhat automated as well), and (very importantly!) the fact checker for the output.

On the promotional side, AI tools will also make writing book blurbs, ad copy, etc. easier for novices. It will allow for creating more variations that can be tested, to see which of several ads sells the most.

Stay tuned for more information!