Last week in the Avid Readers of Christian Fiction Facebook group, someone asked for opinions on AI use in Christian fiction. And, as always when it comes to this subject, opinions were fierce and many.

I wrote about using AI as a tool way back when it first stormed onto the scene, and I still stick by my sentiments expressed there, generally speaking. But when I wrote that, I had honestly not experimented with AI at all beyond image generation–personally. My husband was already using it every day to help write code for websites and software and the like.

In the intervening years, I have experimented with it in some ways, so I can speak more knowledgeably. And what’s more, I’m seeing SO MANY authors put out statements…often to the effect of “I will NEVER use AI in any way, shape or form.” And I respect the lines they draw. I don’t necessarily share all their lines. But I definitely have my own, and so I wanted to talk through them. Because, as with most things in my life, they’re a bit too nuanced for a social media meme. 😉

1. Will I ever use AI to WRITE a book or other commercial project?

Absolutely not. Ever.

I could leave it at that, because this is a very cut-and-dried line for me. But I want to talk about it anyway, so I’m gonna, because this is my blog post and I can write about it if I want to. 😉

So here’s the thing–I’ve used AI to write some simple website copy. Specifically, when I was doing the “tour” of Alnwick for the Imposters page. These were two-to-three sentence descriptions of real locations, so I had ChatGPT write them, and I edited them. It saved me an hour or two of work, and this is not something that I consider “defining” of me, it’s not something I’m selling, and frankly, I could take down the page and no one would likely even care, so…I see no harm.

But beyond that? Nope. Not blog posts, not articles, not books, not…well, that’s really all I write, but if ever anything else came up that was part of my body of work, I’d say no there too.

Because writing is WHAT I DO. Writing is WHAT I LOVE. Writing is the FUN part. Why would I give that up?

I wouldn’t.

And you know the funny (and annoying) thing? I’m getting pretty darn good at picking out AI-written content, and I bet you are too. And it annoys me, because the phrases and uses that always clue me in are perfectly legitimate phrases and uses, and I’ve habitually used them too. (It was trained on us, after all!). Things like em-dashes and ellipses and “that isn’t____–that’s ____.” (That construction is in EVERY SINGLE AI ARTICLE I see on social media, and it drives me NUTS.) Which means that when I actually write something that sounds a bit AI, I grit my teeth and edit it out, even though it was 100% me. Because I don’t want it to sound like AI, even though AI is just sounding like me there.

See? Funny. And frustrating. Though for the record, em-dashes and ellipses will STAY in my writing. Because I love them. They’re mine. They don’t belong to AI. And I stand with the many other authors who say, “I was using them well before AI began to, and I’m not going to stop now. They’re part of my style and voice.”

And while we’re on the topic…

2. AI Detectors

I am dubious of AI Detectors. Why? Because they are definitionally hypocritical, for one thing. “I’m going to use AI to detect whether you are using AI.” Yeah, um…no.

Moreover, I have heard from soooooo many friends where these detectors are laughably wrong…because AI was trained on us. Lots of long-time authors are plugging in their books written well before AI existed into these detectors, and the detectors are returning that they were 30% or 50% or 80% AI…when that just isn’t possible. Again–AI sees itself in their books because AI was trained on their books. So of course there are similarities.

So–another funny story. At WhiteFire, we got a submission that sent all my own AI-detector senses tingling. It felt completely AI-generated to me. But I didn’t want to just make an accusation. So out of curiosity, I plugged a sample into an AI detector, and it came back with 99% AI-written.

This matched my gut instinct. But could I trust this detector? Well, I decided to experiment a bit more. I plugged in an equal amount of The Memory of Freedom, which I was working on at the time. And which I obviously knew was 0% AI written, because I was the one to write the thing. 😉 And it said…0% AI written.

Okay. Sweet. But is one data point enough to convince me it’s accurate? Of course not! So I opened up another of my manuscripts, and then another, and plugged those in too. They all came back 0% AI, which they were. So, you know. That made my day for some stupid reason, LOL. It’s not like I didn’t know that they’re all me, but even so, I’d been seeing so many of my author friends alarmed because when they did the same, detectors were spitting out higher numbers, even though they hadn’t used AI on them, and they were afraid they were going to get wrongfully accused by readers or publishers.

Which is a real thing, friends. People are tossing around these accusations with self-righteousness and a total disregard for how it can impact an author, and not taking into consideration that thing I keep pointing out–that sometimes AI sounds like us because it was trained to.

But I digress. Let’s move on.

3. What about using AI to brainstorm?

I admit it. I was dubious about whether AI could be a good brainstorming partner. But I actually really love using it as one. And here’s why.

For most of us creatives, part of the brainstorming process is just putting things out there. Writing it out. Getting ideas down, throwing them against a wall, and then seeing what sticks after you spin that whole wall around a few times. I have countless documents for each project that are just me writing to the screen as if I’m writing to my best friend/critique partner. They are full of “maybe I should…” and “or, no, I don’t like. Maybe I’ll…”

They’re a mess. But it’s part of the process. Sometimes I’ll send these thoughts to said BFF/CP. Or to my husband, or my editor. And they’ll respond, and I’ll think about what they say…and then I’ll totally ignore them and go off in some third (or eighteenth, LOL) direction that has nothing to do with anything we talked about but which was sparked by a single word or phrase. (David laughingly complains about this, saying, “Why do you even bother asking me when you ignore every single thing I say?” I keep telling him I don’t ignore him, he just inspires a whole new direction, but you know…LOL.)

It’s fun to use AI as this sounding board. Because it will come back with ideas that are…honestly, it doesn’t matter how they are. Sometimes they’re decent. Sometimes they’re laughably horrible. Never do I actually use anything it says in the form in which it says it, but just like with myself or Stephanie or David or Janelle or Elizabeth or whoever-else, the process itself gets the juices flowing. The back-and-forth creates the movement in me that I need.

And sometimes, its vast knowledge is indeed helpful, especially as I’m in that stretching stage of an initial brainstorm, where I just don’t have facts enough yet. I can ask it questions like, “Does it make sense to have this setup in 1943 France, or would that not be plausible?” and it will give me an answer.

Now, one place I really LOVED what AI gave me was when I was developing the “mythology” of my fantasy world. I wanted it grounded in real mythology, but with my own unique twist…and you know what those databases have all of? Mythology. Again, I didn’t use most of what it suggested, but its input gave me just enough to get excited and take off on my own, using as a base some of the real mythology of our ancient societies.

But then, that really gets more into…

4. AI for research?

Yes. I love AI for research. Is it imperfect? Absolutely. Do I need to fact-check it? You bet. But here’s why I like it even more than a Google search, which I also adore:

I can ask it questions without knowing the right keywords to yield the results I need.

And as someone writing a lot of books set in France right now, I can also ask it for information that isn’t available in English. This is so helpful. One of the first things I used it for like this was actually the name of the building the Gestapo was housed in in Berlin. I could NOT find this in English for some reason. I spent a ridiculous amount of time searching the internet, got frustrated, asked ChatGPT, it spat out the answer in 3 seconds, I then put what it gave me into my search engine to verify, and boom. There it was.

Similarly, I can ask it things like, “Can you give me a list of 20 French female names popular in this specific region of France in 1920?” SO HELPFUL. Otherwise I’m cross-checking “popularity” lists with “location” lists, and it takes so much longer.

5. AI for “fill in the blanks”

Okay, this is my own need right now. But it’s been…huge.

As you likely know if you’ve been hanging around here any length of time, I had chemo in 2024. And I’m having chemo again now–all year. During both that initial, intense schedule of infusions and certainly this year with what we could call “chemo-light,” I’m still writing. I have books due. Contracts, deadlines. No time off for things like cancer in my life!

And chemo brain? It’s real, friends. I’ve been so, so blessed not to experience it to the extent that so many people do…but it’s still there.

It hits in the strangest ways. One I’ve noticed lately is this “out of time” feeling. I’ll be doing something random, like walking through the living room, and I’ll have this strange feeling of “What season is it?” It’ll take me a second to process what month we’re in and what that means seasonally. Which I mention solely because it’s weird, LOL, and indicative. But here’s where the brain condition is relevant to this conversation:

I forget words sometimes. I mean, we all do that, it’s nothing new. But it’s definitely happening more than it used to. Not enough to be alarming–but definitely enough to be frustrating.

Because I’ll be writing, going along at a great clip, in a groove, and then BAM. I hit a wall where I cannot for the life of me think of the word that I KNOW, and which is on the tip of my tongue, but which I canNOT pull forward.

You know how that feels–again, we all have those moments.

Well, when I have them in conversation, it remains annoying. But when I have them in writing…I don’t have to get frustrated anymore. I can simply copy and paste that sentence into ChatGPT with ____ in place of the missing word and say “I can’t think of the word I need. How should I fill in this blank?”

And it gives me options. Inevitably–every single time–the one I’m looking for is its first or second recommendation.

Guys, this is HUGE. As a writer, words are my lifeblood. And to feel like sometimes they’re evading me? That’s rough. Before it occurred to me to use AI for this purpose, it was not only frustrating but disheartening to keep running into these ridiculous problems. I would sit there for minutes on end trying to find the word. And the longer I searched for it, the more frustrated I got. Which of course just made it worse.

Now, seconds. No emotional spiral. I can find the word and move on.

6. Isn’t it stealing?

Another concern I saw in that post last week was about whether AI is using stolen work.

This has been examined by the courts already, and what they decided was that training AI on our existing works is within the legal scope of fair use–because teaching has always been a fair use of existing works. What was not okay was the fact that Anthropic used pirated copies of the books it trained on. Which is why they now owe lots of money to all of us whose pirated books were used.

What does this mean? That according to the courts of the United States, training is okay, as long as the works being used to train are either free or paid for.

This perfectly aligns with my own understanding of copyright law, so I wasn’t surprised to see this verdict come down, and it also answers the objections to my mind. Because AI does not regurgitate the books it was fed. It might write something “in the style of” a very popular author that will have a lot in common with existing books by that author if you prompt it correctly, but it will not deliver you the text of that book.

It will quote parts of public domain works for you (edited to add a friend pointed out it will also quote passages of copyrighted works; so far as I know, this is only brief quotations, not full chapters or the whole book. Traditionally, brief quotations have fallen under “fair use,” though it’s certainly something to keep an eye on); and it will also (currently) yield search results that can include images, which are linked and cited, much like any search engine does. And as with any search engine results, you the user must use this legally.

This is no more questionable than any internet search; which, let it be noted, will also yield plenty of pirated versions of things. (Traditional internet searches will also yield quotations from copyrighted books; I have many times been researching a topic and end up with a search result from a passage of a book that’s in Google Books. They show me a few paragraphs–not the whole book.)

Much responsibility ultimately lies with the end-user, as always. Which is another reason why I have the lines I do–I always use other people’s work for research and inspiration. I never use their words and claim them as my own.

7. AI for images

I love AI for images–I’ve not been shy about that. I love generating character images, scene images, you name it. As someone who has Photoshopped thousands of images over the years, being able to generate what I want is so much fun and so much quicker than painstakingly combining a dozen images to get the one I want.

Now, it’s often (always? LOL) imperfect. I’m still going in and making adjustments.

And if a client of mine (keeping in mind I’m a cover designer) is anti-AI, that’s totally cool. I know how to work without it. No biggie.

As with stock imagery, an AI image all by itself is not MY (or the author’s) property. But when I add my own work to it, it becomes my own work–that’s where the law came down. And this is always what I do. Much like with stock imagery, I combine a bunch of different elements to make a cover unique–that’s why people hire me.

I’ve heard of cover designers losing work because of AI…but honestly, I haven’t. My schedule is still booked out as far as it’s ever been. Because a good designer does more than slap words on a picture.

Now–I get that we want to continue supporting human artists. And I do. I still have my Shutterstock account, and I still download a gazillion images by real human people, even when I’m using generated images too.

I have one client who wrote “No” on the question of whether they were okay with AI, so I ended up using something like 14 different images on her cover. But the lighting on some of those images was wrong, and all my fiddling with it wasn’t yielding the results I wanted. So I took the image I had created with about 12 of those 14 images and asked ChatGPT to even out the lighting. Voila! It did. I then showed it to the author and asked her if she was okay with that particular use, assuring her that I paid 14 different human artists already, LOL. And she was totally fine with it. Because the AI didn’t create the image, it edited the image. Which is, frankly, something my software has been doing for decades already, AI just did it a little differently (and in this particular case, better than I was managing it).

When it comes to images for myself though, I love it because it gives me a chance to have images that I otherwise just wouldn’t have. Especially in the fantasy world, but in historicals too. Because they just don’t exist in photography, and to create them like I do book covers would take hours I can’t justify spending on such trivialities. But having those visuals is so cool.

8. What about environmental impact?

This came up frequently in that Avid Readers post, and I absolutely agree that the environmental impact is real and needs addressed.

And honestly, I think it will be. Why do I think so?

History.

Look at the impact of the industrial revolution. Electricity. Automobiles. All of these advances came with what turned out to be environmental impacts. And because they were popular and in demand, those issues were worked out.

I imagine plenty of people said, “No factories! They’re hurting the environment! No cars–the fumes! No electricity–look what damming the rivers for the power plants does!” And they’re not wrong…but most of us today are still using all our factory-made goods, driving our cars, and running our houses on power. Because the issues caused by these technologies have been dealt with (some better than others, obviously, I’m not claiming there aren’t plenty of issues still) or compensated for or otherwise deemed “worth it.”

Yes, AI is absolutely consuming a ridiculous amount of energy and resources. This needs to be addressed. But I don’t think the answer is “cut it off and don’t use it.” I think because people are using it, the answers will be forthcoming and in short order. When there is demand, there is innovation and resolution.

9. In conclusion

These are obviously all just my opinions. Why I like what I like and why I don’t like what I don’t like.

I absolutely respect different opinions. All those authors who will never use it all for anything? I have total respect for that. Just like I respect those who hand-write as part of their process, or those who use dictation, or those who write out-of-order, or any other process different from mine.

But I don’t think it’s a moral thing, as long as we’re using AI ethically. For me, that means it does NOT ever write for me, but it’s okay to use it as a tool along the way, much like I use plenty of other tools. The real “magic” of a story always still happens in my own brain, not in its algorithm. And the “magic” of crafting the words is a joy I will never give up, thank you very much. But as a shortcut for research, or acting as the wall against which I throw my “spaghetti”? I personally see no problem with that.

I can promise you that everything you ever read by me is written by me, created by me, developed by me. That has always involved outside sources–books, articles, websites, magazines, blog posts, videos, podcasts, documentaries, you name it.  But my stories aren’t created by those sources. They’re just input. Data. It all goes in, gets filtered through the wacky mind that is Roseanna, and then gets turned into something else.

A story, through the eyes of characters who come to life for me.

The sources of the data always change. From book to book, year to year, mood to mood, frankly. Sometimes, now, some of that reading is of AI results. But it’s no different, to me, than one more article. Just data to put into the Roseanna Brain. One more source. At least, that’s how I view it.

What are your thoughts? I’d love to hear them.