I don't use generative AI to write

Using generative AI in my writing process does not interest me

Because you do not live beneath a WiFi-impermeable rock, you know that people are having feelings about the place of generative AI in the creative writing process (assuming there can ever be a place for it at all); and moreover, they have been having these feelings for quite some time.

I myself have feelings. However, I've noticed that, historically, other people are rarely (if ever) interested in my reactions to or input on what they're doing, so rather than uselessly expend my effort here on telling the world what I think about its generative AI usage, I'm just gonna put down how I feel about using it myself, as follows:

I'm not interested in it.

I like writing. I find the process pleasurable. The actual act of typing words, one character keystroke at a time, as I preview a reality via a brainstorm or simultaneously imagine a person or place or conversation, is exciting, because I get to both witness all the action unfold and influence how it will go. The act of making words with a pen in my hand is exciting, too, because I write longhand more slowly than I type, and when I work longhand (which I do, sometimes--that's how I drafted my latest sale), I have time to feel and imagine the atmosphere and setting even more deeply.

When I'm writing, I'm not here. I'm someplace different and wholly new, a place that nobody else has ever been, and the only way to have that experience is to actually have the experience. The piece of fiction that results from engaging in the practice is, in a way, merely a keepsake or artifact of the actual event, like a series of photos I've taken while on a trail run.

But the photos are not the run. Asking a computer to create images (even if they wind up being gorgeous, uplifting, compelling images) of a trail does not also give me a workout, and it never will. 

I want both the resulting keepsake and the experience of the event. But generative AI and LLMs, no matter how realistic their outputs might become, are only capable of giving me keepsakes from a non-existent experience. 

I know that not everyone is like this. Plenty of people buy and wear things that advertise places they've never been or universities they've never attended, so there must also be plenty of people who actually hate writing but want other people to think that they have done it.

I'm just not one of them.

One time I used an AI image generator for a book cover (and I was not impressed)

I have never used, or even been curious about using, generative AI or LLMs at any point in my own writing process.

But, in the interest of encouraging honesty and transparency in these types of discussions among creators, I will say that, one time in early 2023, I did use an AI image generator to generate a cover image for one of my self-published books, a small collection of erotica I'd written early in my career called HEART OF BRASS AND OTHER FANTASIES.

As I was putting together HEART OF BRASS (in mid-January of 2023), I went to Shutterstock to pick a stock image for the book's cover. At the time, Shutterstock was promoting its nascent in-house AI Image Generator, and it allowed users to submit a certain number of prompts a day for free and download the resulting images, which would be capped at a certain resolution and limited to a certain canvas shape.

Also at the time, I had started to hear some discourse surrounding the ethics of AI usage (namely regarding the intellectual property theft part), and because of these concerns, I felt leery of trying any generative AI tools at all. But Shutterstock claimed that its AI Image Generator had only been trained on stock images from its own image library whose creators had expressly agreed to the usage of their work in this exact way, and I thought, "Well, maybe it would be ethically okay to try this particular image generator, then."

Looking back, I don't know if Shutterstock's claims about the permissions it received for its image generator's training data were actually true. (Here in 2026, we know that people who gather training data to feed to these AI models often source it problematically). But what's definitely true is that using Shutterstock's AI in 2023 to create the cover image I was hoping for was disappointing and frustrating.

Maybe I was expecting too much of an early image generator. (I first prompted it to create an image of a woman leading a dragon on a leash; the "woman" was nightmare fuel and the "dragon" was some bizarre mashup of a legless wolf and an ink stain.) Maybe my prompts should have been more specific. (I tried focusing on just body parts next, like "a dragon's tail" or "a woman's legs from the knees down, wearing thigh-high boots" but was also met with Yikes.) Or maybe I should have aimed even lower to begin with and started out with prompts for objects instead of subjects. (Nevertheless, the AI failed to convincingly render high heels, gold chains, bondage cuffs, collars, handcuffs, whips, floggers, pentagrams, angel wings, peacock feathers, leaves, steampunk goggles, pocket watches, and swords.) 

When it finally created an image that I thought was halfway acceptable (with a masquerade mask, dagger, and rose), I still had to work with the image in Photoshop to remove nonsensical artifacts (like extra stems on the rose, a jagged part of the otherwise smooth dagger blade, or irregular hints of other objects in the composition). And the end result, if you zoom in at all, still looks kinda jank.

What's my takeaway? My experience using generative AI was "ugh," my final cover is "meh," and, even assuming that the same prompts today would elicit better images from modern AI models, I've learned a lot more since 2023 about the ethical concerns around AI usage (including the fact that, since many companies lie about the origins of their AI model's training data, I was probably naive in trusting Shutterstuck when they said that their own training data was issue-free). Using AI for book cover generation is not something that I would do again.

In the meantime, for now, my cover for HEART OF BRASS stands. I recently updated the book's metadata in Amazon to indicate that I used generative AI in the creation of one or more of its images (a metadata field that was not available when I first published the book in 2023). One day, perhaps Amazon will be fully transparent with its shoppers about what works and products on its platform were created in whole or in part with AI. So if HEART OF BRASS doesn't have a new, AI-free cover by then, don't be surprised if the book is marked as such in some way. 

Q: Why don't you replace the cover of HEART OF BRASS?

A: There's nothing that says I won't.

Q: I mean, now that more and more creative voices are agreeing that AI Is Bad, Actually, why isn't changing this cover image moving higher on your priority list?

A: There's nothing that says it isn't. That priority just hasn't yet moved high enough to supersede all my other creative priorities, mostly because (1) I'm not a popular, highly visible creator whose choices have a wide impact and therefore outsized potential for harm; (2) the book has already been out for over 3 years, and in that time, I've received zero commentary (publicly or privately) about the cover; and (3) while Current Me would never create a cover with AI, this is something that Past Me did, and if I just quietly changed the cover one day without first making some type of post as to why I had done so, it would feel wrong to me. Even if you can hide a past mistake from others, you can't ever really hide it from yourself. Better to not hide it from anyone and talk about it so that everyone can learn something.

Q: That last one is a nice sentiment and all, but it doesn't quite logically follow from my question.

A: I mean to say that posting about my experience using an AI to generate a book cover image is/was a higher priority for me than immediately changing the cover image itself. So people can use this new knowledge and go witness the cover in question for a while, if they want. 

Q: You could have just updated the book's cover image and then put its original cover image in a mea-culpa-type post so that people could still examine the image without it being associated with your book. You know that, right?

A: This interview is over.