Let's Talk About AI

I want to start by stressing that I NEVER, nor WILL I EVER, have AI write any part of any novel or screenplay I create. To me, that would be just like having another person write a part of my novel and taking credit for their work. On the other hand, I don’t really care personally if other people do it. If someone thinks they can program a robot to write better than I can, bring ‘em on. As far as an AI being “trained” on my work? Well, we’re all trained on the work of other writers. If an AI dubs my work worthy of training purposes, I’d actually be kind of flattered.

It seems to me that visual artists feel differently—that every image or video created by an AI program is an affront to most of them. I can’t speak to that—I’m not a visual artist (we’ll get to that in a second), but I can see why they would feel that way. We writers are used to everyone, no matter how unskilled, thinking they can do what we can do. For visual artists, it’s a new experience.

I do use AI for three things. Let’s start with the most controversial. As I mentioned, I’m not a visual artist (told you). But The Book of Monsters requires pictures. The appendix is basically a monster encyclopedia, and you really can’t have a monster encyclopedia without images of the monsters being described. I can’t draw them, and I can’t afford to pay an artist to create all the images that would be required, so I made ‘em with AI.

I think I’m still on solid ground ethically because A) It’s not like I’m profiting from these images and B) My intent, once the book is sold, is to work with the publisher to engage the services of a human illustrator to recreate the images using the AI versions as a guide. I’m just using it as a tool—which is what it’s supposed to be, I think.

(As an aside, I’ve been told that the use of em-dashes is a bit of a fingerprint for AI writing, so I’m really self-conscious about using them, and they always jump out at me now. But I promise—this is all me.)

Anyway, AI-created images are commonly referred to as “AI slop,” but I think the ones I generated look pretty good. I’ve included a few at the end so you can judge for yourself.

Moving on. The second thing I’ve used AI for is to find agents to query. I’m not afraid of the brute force approach—as I mentioned, I queried 100 agents the hard way. But having an AI scour the internet to try to find the agents who are looking for EXACTLY or almost exactly what I have to offer, well, this seems to me to be just the type of thing AI was intended for.

The last thing I used it for is help with my query letter. Basically, now that my AI friend knows exactly who we’re looking for and why, I want it to use that information to help me build a query letter that’s going to tickle the dopamine receptors of those specific people. That makes sense, right?

Now, I wouldn’t have any real ethical problem with sending out a query letter created wholly by AI—the query letter is a tool, I’m not making any money from it. But you can’t really do that, and I think it’s important to talk about why.

As I stated, it’s not because of any ethical reason. It’s because no matter how much information there is about query letters on the internet, the AI doesn’t know how to write a query letter about my book, even though it’s read my book. And to understand the reason for that, it’s really important that we understand what AI is.

We call it Artificial Intelligence, but it’s not that, any more than Hoverstar hoverboards resemble the hoverboards of 2016 in Back to the Future II. AI is just a sexier name for LLM—Large Language Models. And what that really means is that your fancy AI is just a high-powered Google that’s great at research and has learned to communicate in English.

Now, to those of us who grew up with Text-based adventure computer games that could only understand two-word commands like GET KEY and LIFT GATE, this is the Holy Grail of technology. But it’s not something that is capable of independent thought, like Data or HAL 9000. Not even close. Ask it. It’ll tell you I’m right.

So why do they call it AI, other than that it feels more science-fiction-y than “Large Language Model?” Well, because there is one thing it CAN do, at least the good ones, that IS like those sci-fi AIs—it can learn.

So if I tried to send the first query letter AI spit out to an agent, they would probably be very confused. It probably wouldn’t make a lot of sense. It would be a collection of heavily researched rules and tips mashed up with information about my book to produce something that looks a lot like, but is not exactly like, a query letter about my book. But I can look at that query letter and say anything from “Hey, you got the word count wrong, it’s this,” to “Hey, this comes across as a little too self-congratulatory, we want to be more humble,” or “It will make more sense if we talk about the sequence of events in this way, not that way.”

Even better, when I get feedback from an agent who has rejected me ON that query letter, I can go back to AI and say, “hey, Agent X suggested that more of the plot be made clear in the query letter,” or “Agent Y said a query letter needs a hot opening line to really grab the reader,” and it will listen and try again, and give me something better that I can build on when I write my next query letter and send it out.

The long and the short of it is, I think a lot of people feel that AI is the enemy of the artist. And I can see how it could be. But I think, if used properly, it can be an invaluable tool, and that’s how I’m going to keep trying to use it.

Weekly Tab: 5 queries, 1 rejection

Vampire, Romantic Type Petrifier, Gorgon Type Phantasm, Yurei Type