The issue of artificial intelligence (AI) and large language models (LLMs, like Chat GPT) continues to cause turmoil in the writing and publishing world. There was recently a writer profiled in an article who claims to write hundreds of books a year, published under a variety of names, using AI, though it doesn’t seem as though any of them sell very well and the way she makes her money is by teaching other people how to generate books using AI and selling AI prompts and even complete AI books that the buyer can edit and customize. There have been authors caught leaving in the AI prompts, so it seems they used AI to write at least part of their books and just pasted in the result, and somehow no one actually proofed the final book to notice that the prompt was included in the pasted passage.

On the other hand, there seems to be a bit of an AI witch hunt going on among some reviewers, with hours-long YouTube videos analyzing books as to whether or not they may have been AI-generated or reviews calling out books for having hallmarks of AI — even though they were published before AI existed. Sometimes it’s more a case of AI imitating existing books and writers’ styles, since that’s what they were trained on. I have a number of my books included in one of the lawsuits against an AI company, since they trained their AI on pirated books. That means there’s a chance some AI writing could sound like me — and vice versa.

What some artists are doing to counter possible accusations of AI use is documenting their process. If they get accused of having used AI to generate a piece of art, they can show the pencil sketches and the various phases of the work. Authors are starting to do the same thing, documenting and discussing the process so that if someone accuses them of using AI, they can show how the book developed. The reviewers on the AI “witch hunt” do seem to check authors’ social media and other writings to see if there’s been any discussion of how the author feels about AI and of the process of creating a book. I suppose you could create AI prompts based on the kind of work authors can talk about, since we can’t really show the pencil sketches the way an artist can unless we publish every draft of a book, but showing that there was some thought that went into the writing can help counter some accusations, or at least shed doubt on them.

The tricky thing for me is that I don’t really like talking about a book while I’m writing it. A book in progress is a really fragile thing for me. To use a metaphor from my recent pottery class, it’s a lot like when you’re shaping something on the pottery wheel. If the thing you’re making gets jolted in any way, if your hands move in the wrong way, what you’re planning to be a vase may end up being a pitcher (true story). I’ve found that I can’t even talk in specifics to my agent. I thought it would save me a lot of time to hash out the idea with my agent before I start writing so that I know I’m on the right track, but when I did that and had an idea she approved of, I found that I couldn’t write the book. I didn’t care about that story anymore and when I tried writing it, it came out dull and lifeless. It wasn’t really my story anymore because someone else had been a part of shaping it. I may brainstorm with someone when I’m well into a book and need to work out a plot twist or an ending, but the book has to have taken shape by then. I can’t talk about it in the idea stage.

There’s also the issue of spoilers. If I talk too much about a book, I might give readers more info than they want before they get to read the book for themselves. Or it could make people think they’ve already read it. I could probably give more specifics in my newsletter because those are my most devoted fans who will be buying the book based on my name or the series. But if someone’s just scrolling through social media and sees something without registering who it is, there’s a chance that if they ever come across the book they might think it sounds familiar and figure they already read it.

And there’s what I think of as the Rogue One trailer effect. Between the time the first trailer for the movie Rogue One was released and the time the movie was released, the movie had been rewritten and in some cases even re-shot, so most of the scenes that were in the trailer weren’t in the final movie. There were even bits in the trailer that were never meant to be in the movie. The camera crew wanted to make the most of some of the locations they had for a limited time, so they shot things that weren’t in the script if they saw something that looked cool, and those cool bits were used in the trailer. The final movie was great, but the first time I saw it, it was disconcerting because I kept waiting for the trailer moments that never came. My books go through a lot of changes in the revision process, so there’s a good chance that if I talk about a scene I’m working on, that scene either won’t be in the final book or will be drastically different in the final book.

But I may try to document some of my process in a general sense without talking in specifics about the book. That can help readers know the kind of thought that goes into a book, and I hope that will show anyone wondering about AI use that my books are entirely artisan and organic. I’m in the development stage of a book so may have to catch up, but otherwise I’m starting at the beginning of a project.

For the record, I don’t knowingly use AI for anything. I feel like using AI to write would be like taking a forklift to the gym. Yeah, I could make it lift huge amounts of weight, but I wouldn’t get anything out of it, and it would probably cause damage to the gym. I write to get stories out of my head, and I can’t see how feeding a prompt to AI would do that for me. The story would still be in my head, and the result wouldn’t be the story in my head. I wouldn’t become a better writer by having a machine churn things out for me. I don’t trust it for research because it doesn’t give you facts. It gives you something that sounds like a report of facts, and the facts may or may not be true. I can’t imagine using it for brainstorming because it’s giving you the synthesis of what’s already been done. You’re going to get boring, generic ideas. And then there are the ethical issues. They were trained without permission on work stolen from other people (including me) and they have a huge environmental cost. The reason I say “knowingly” is because they keep trying to add AI to everything. I do use spellcheck, and while that technology has been around for a long time, they’ve started using AI for that (and making it worse). I’m not great at spotting AI art, so while I try to use artists and designers who avoid AI, people can lie and I might not know. I try to track images I use for promo that come from the image library of the applications I use back to the source to see if they were AI-generated, but it’s hard to tell sometimes and I may miss something.