I recognize the irony in opening this page with that quote, considering Marie Curie's pursuit of understanding ultimately exposed her to the very radiation that likely caused her death.
And yet, here we are.
Humanity continues to move forward. We invent things. We improve them. We find new ways of doing old things, and then spend an extraordinary amount of time worrying that those new ways are going to destroy everything that came before.
Technology has always changed human civilization, of that there is no doubt. It has created entire industries, transformed others, and destroyed more than a few along the way.
I should know. I've lived through more than one of those transformations.
The Death of Graphic Design
I was a graphic designer when Photoshop first arrived, and I remember the predictions of doom from my fellow designers.
“Graphic design is finished.” “The profession is dead.” “Any fool with a computer can now do the work that people like me have spent years learning how to do.”
And to be fair, there was some truth to their fears. Photoshop fundamentally changed the graphic design industry. Work that once required several people could suddenly be accomplished by one. Skills that had taken years to develop became less valuable. Some jobs disappeared. Some businesses went under. But graphic design didn't die.
Instead, designers could suddenly do things that had previously been difficult, expensive, or downright impossible. The work became more dynamic. The possibilities expanded. And because the cost of producing that work dropped, more people could afford to buy it.
The industry didn't disappear. It evolved. And that was hardly the first time something like that had happened.
There was a time when a chandler, the person who made candles, was an enormously important member of society. Candles provided light, and light was something you needed if you wanted to do anything once the sun went down. Chandlers could be wealthy, respected, and influential members of their communities.
Then came kerosene lighting. And the world changed.
The automobile devastated the horse-drawn carriage industry. Digital photography transformed the film industry. Electric lighting replaced kerosene lamps, and the technology continued to evolve, from incandescent bulbs to compact fluorescents to today's LEDs.
Each advancement changed the industries that came before it. Sometimes with devastating consequences for those whose livelihoods depended on the older technology.
But that's what technology does. It changes things. And AI is no different.
Is AI Going to Kill Us All?
There are people out there predicting that AI will destroy humanity within the next few years.
Maybe it will. Who knows?
My question is: will humanity even survive long enough to give AI the opportunity to destroy us? Or will we manage to kill each other over some monumentally stupid difference of opinion first?
Personally, I think rumors of humanity's impending doom have been grossly exaggerated. We've been predicting the end of the world for thousands of years, and yet here we are, still arguing about it.
Now, I'm not saying AI is harmless. Far from it. AI is a powerful tool, and powerful tools can be used to accomplish terrible things. Words are tools, and history has repeatedly demonstrated just how much harm can be accomplished with the right words in the wrong hands.
AI can amplify that power enormously. But the technology itself isn’t inherently good or evil, no more than a hammer is. A person can use a hammer to build a shelter for someone in need, or bash that someone’s head in.
It’s the person wielding the tool that brings the moral compass. And whether you like it or not, the tool that is AI now exists.
Pandora's box has been opened. We can debate whether it should have been opened in the first place, but we’ve lost the ability to put this bastard back inside.
So rather than spend my time fearing it, I would prefer to understand it. Which, I suppose, brings us back to Marie Curie and what this page is supposed to address.
AI and the Written Word
Let me begin with something that will probably upset a few people.
As a writer, I do not fear AI. Not even a little.
AI can write a beautiful sentence. Sometimes. It can even write a beautiful paragraph. Sometimes. It can even put together an entire scene that, taken by itself, is wonderfully written. Sometimes.
But there's a problem. A beautifully written scene can absolutely destroy a novel.
Imagine a scene that contains wonderful prose, believable dialogue, compelling emotion, and a fantastic character moment. On its own, it might be brilliant. But what if that scene reveals something the reader isn't supposed to discover until later? What if a character makes a decision that seems perfectly reasonable in that moment, but undermines their emotional transformation? What if that wonderful scene destroys everything the story has been building toward?
And that is one of the fundamental limitations I see in AI-generated writing today. A novel is not merely a collection of well-written sentences, paragraphs, or scenes. It’s character arcs, story arcs, plot arcs, thematic arcs, symbolism, subtext, allegories, foreshadowing, misdirection, emotional development, and hundreds of other decisions whose significance may not become apparent until hundreds of pages later.
And all of those things must mesh together to create a work of art.
Today’s AI can remember an extraordinary amount of information. But remembering information is not the same thing as consistently understanding how every part of that information affects every other part.
Consider this: You ask AI to create an illustration of a long medieval hallway. To the layperson, that hallway might look gorgeous. Beautiful stonework. Dramatic lighting. Intricate architectural details.
But then you notice something odd—the wall lamps. One is too high. Another is too low. One is centered on its wall while another is practically hanging off the edge of a pillar. Half of them don't even look the same.
Why? Because AI knows what a wall lamp looks like. It knows what a medieval hallway looks like. But it doesn't reliably understand the architectural consistency that should govern how those things fit together. It has produced an attractive image without maintaining a coherent underlying design.
And AI storytelling suffers from precisely the same problem.
And nowhere is that failure more apparent to me than in subtext.
Not merely placing hidden meanings into a story, but understanding why those meanings matter. Understanding what a character believes, what they refuse to admit, what the reader knows that the character does not, and how those layers interact with everything that has come before and will come after.
AI can imitate those things. Sometimes quite convincingly. For a short moment, anyway. But it doesn't have the human experience behind them. It doesn't feel grief, love, jealousy, shame, or triumph. It can describe those emotions, and it can recognize how human beings express them, but that is not the same as experiencing them.
Now, does that mean AI can't write better than some published authors?
Oh, hell no. There are plenty of people publishing stories on the internet whose writing is considerably worse than what AI can produce.
(Of course, I'm not talking about you, my friend. Your stuff is amazing. We both know who I'm talking about, though. Those writers over there. Hacks, at best.)
But that brings me to the next thing that will probably upset a few people.
Yes, I Use AI
And to those who find that statement offensive, I say: Pshaw.
I also use Photoshop. These are simply two tools of many (MS Word being another, and Illustrator, Premiere Pro, InDesign, Firefox, Thunderbird, OBS, GitHub, FontForge, my Wacom tablet… the list goes on) that I use to increase my productivity.
And AI is no different. I use it to talk through ideas. I use it to examine grammatical problems. I use it to challenge my thinking, explore possibilities, and occasionally point out something I've overlooked. Not to mention that it is an absolutely magnificent thesaurus.
For those of you who haven’t rage-closed this website, let me explain.
Decades ago, I would write a sentence and realize that one of the words was weak. Not necessarily wrong, but not quite the word I wanted. However, try as I might, I simply couldn’t conjure up the right word in my mind as I sat there staring at the sentence. This has happened to every writer since the dawn of prose.
And like all of those who came before me, I would pull my trusted thesaurus off my bookshelf, then I would spend twenty or thirty minutes reading through lists of words, searching for the one that expressed precisely what I was trying to say.
Years later, I began using an online thesaurus called OneLook. I had the same problem, and the online thesaurus gave me the same solution. But OneLook allowed me to narrow my search much more efficiently. Instead of spending thirty minutes searching through a physical book, I might spend ten minutes looking through a more focused list of possibilities. Then I would choose a word and continue writing.
Today, I use AI for the exact same thing. I write a sentence. I identify the word I'm unhappy with. Then I explain what I'm trying to accomplish and ask AI to give me alternatives. Seconds later, I have a list of possibilities I can look through. And then I choose the word.
AI doesn't choose the word for me. I do. Just as I chose the word when I used OneLook. Just as I chose the word when I used a physical thesaurus.
In all three examples, I wrote the sentence. I decided the word wasn't doing what I wanted it to do. I examined a list of alternatives. And I made the artistic decision about which word belonged in my work. The only meaningful difference is how quickly I can find the word I'm searching for.
Where, oh loather of AI, is the ethical difference? Because, personally, I don't see one.
Here’s the reality: using AI to generate alternatives for a word is not surrendering my authorship any more than using a thesaurus. Asking AI to identify a grammatical problem is not surrendering my authorship any more than asking an editor to do the same. (And don’t you dare spout off the bullshit, “But the editor gets paid!” when you sit there allowing Microsoft Word to correct your own damn spelling mistakes! Hypocrite.) Anyway... Spending forty minutes discussing whether a character would plausibly make a particular decision with AI is not surrendering authorship any more than spending that same amount of time talking to my friend Charlie about the same thing would be.
I am still the one making the decisions. The machine is simply increasing the number of possibilities I can examine in the time I have available. And that means I can spend more time doing what actually matters.
Writing.
And that, to me, is exactly what a good tool should do.
AI-Generated Artwork
Of everything I use AI for, generating artwork is probably the most common. And while this is definitely the most controversial of AI uses, I have some very specific feelings about it.
First, I do not believe AI-generated artwork has replaced human artists. It can create some extraordinarily beautiful images, sure. And the technology has improved dramatically, sometimes resulting in something genuinely breathtaking.
But there’s a difference between creating something that looks beautiful and creating something with deliberate artistic intention. Like a writer, a human artist understands what they are trying to communicate. They make decisions about composition, symbolism, consistency, mood, and meaning. They understand why one detail belongs in the image and another does not.
AI can imitate the results of those decisions. Sometimes exceptionally well. But it doesn't consistently understand the whole.
And to me, there’s another important distinction: the human connection.
When I purchase a painting, illustration, novel, or any other work of art, part of what I value is the knowledge that another human being created it. That person had an idea. They had something they wanted to express. They used their abilities, their experience, and their imagination to create something that I now get to experience. There’s a connection between the artist and the audience, and I value that.
But here's the thing. Not every piece of artwork needs to be a profound expression of the human condition, nor does everything that is written. Sometimes a decorative banner is just a decorative banner. Sometimes I need a picture of a medieval hallway because I think it would look really cool on my website.
So let’s get to the subject at hand and talk about:
My personal policy for AI use
I will use AI-generated artwork for things I would not otherwise pay a human artist to create.
And:
I will not sell AI-generated artwork as an artistic product.
If I am selling you something because of its artistic value, I want that art to have been created by a human artist. I believe that connection matters, and I believe my devotees deserve knowing that is what they are getting.
But if my choice is between using AI artwork for something I would not hire an artist for in the first place or using no artwork at all, then I have no ethical objection to using AI. I wouldn’t pay an artist to create my YouTube thumbnails, decorate my website, or make mockup artwork to help me visualize a monster or character. So I have no issues using AI artwork for those things.
Could I pay an artist to do the above? Sure. But I’m not selling any of it. They are marketing or internal aids only. I used a ton of AI artwork to make this very site, but you aren’t buying this website, nor am I selling any of the artwork I created for it.
But that doesn't mean that AI won't change the art industry. It's already changing it. And just like Photoshop, electric lighting, automobiles, and countless other technological advancements, those changes will create opportunities while eliminating others.
But What About the Artwork AI Was Trained On?
Yes, I know. This is probably the most controversial part of the entire discussion. Generative AI systems were trained using enormous collections of existing material, including work created by human artists and writers. And there are legitimate questions about copyright, consent, compensation, and how that material was obtained and used.
I won't pretend those questions are meaningless. I also won't pretend I have those answers.
But I do believe the discussion is considerably more complicated than simply declaring that AI learned from someone else's artwork and therefore every image it creates is stolen. Human artists learn from other artists. Writers learn from other writers.
I am the writer I am today, in part, because of the authors I read growing up. Robert Jordan, for example, had an enormous influence on me. I never met the man, but I read his work voraciously. I studied how he told stories, how he constructed sentences and paragraphs, and how he developed enormous fictional worlds. In a very real sense, he became one of my mentors without ever knowing I existed.
Brandon Sanderson was also heavily influenced by Jordan, and while our writing styles are certainly not identical, there are similarities in how that influence helped shape our approaches to writing.
Does that mean every book we sell should generate a royalty payment to Robert Jordan's estate? I don't believe so.
Now, human learning and machine training are not identical processes. I'm not claiming they are. But I also don't believe that learning from existing art automatically invalidates whatever comes afterward.
The question of where inspiration ends and exploitation begins is worth discussing. And yes, I recognize that some companies have used questionable, and potentially unethical, methods to acquire the artwork used to train their AI systems. I suspect we'll be discussing that for quite some time.
But as for my own ethical use of AI, I've explained where I've chosen to draw the line. But to reiterate:
I don’t sell AI-generated creative work. Not AI-generated writing. Not AI-generated cover art. Not AI-generated interior art. Yes, I use AI as a writer, but no more than I use MS Word, a thesaurus, my writing group, or discussions with my friends.
The reality is, AI is here. It’s going to change our world, sometimes for the better, sometimes for the worse. But I have no intention of fearing a tool simply because I do not yet understand everything it can do.
I intend to understand it. I intend to use it. And I intend to remain the artist behind the art.
The tools I use may change, but the stories are still mine.