So I went on a bit of a rant on Twitter when I saw a Washington Post article about how writers are already losing their jobs and contract work to ChatGPT. I ended up using a bunch of stuff I hade written a little bit ago when someone in a TTRPG Discord was trying to justify their use of AI art in their project.
It ended up being a whole essay, and the discussion was closed down before I had a chance to post it. I thought I would eventually find time to post it on Cohost, and since I'm still a bit angry I thought why not?
Argument One: AI Is just a tool artists can use
None of these companies making AI art programs or writing programs are making tools. They are making replacements.
They aren’t working with artists, and when they do they aren’t interested in art. How do I know? I worked with one of them.
This was a think tank looking to make something like AI Dungeon. But they were concerned with the black box problem, which is as I understand it, a problem these huge machine learning programs with their huge data sets run into which is: the people who made the software can’t tell why it’s making certain choices. The connections are so vast that it’s just soup, no way to determine why the software weighed one option more than others.
Their solution was to look to TTRPGs. They thought that if they could see a game session play out, and get the thought process from the GM for the decisions they made, this could help them build some software that would make those decisions explicitly.
A friend of mine got in contact, pitched me on this idea, and I agreed. The think tank wanted to find people to provide them with the massive data sets they would need to try out this concept. It would be a solid year of working, recording actual play sessions with me and a player, and getting me to explain the decisions I was making.
Turns out they didn’t want that.
For those with any experience in creative writing and storytelling, you can probably spot the problem. The decision tree made when telling a story can get well, almost infinite. Why is it raining out? Why do you latch on to the fact the rogue had a guild? Why do you offer this or that option? I did my best to record my thoughts and reasoning, and to be succinct, but I inevitably was leaving something out.
The feedback from the think tank? There were too many thoughts. The gold standard they gave from another team pitching them was similar, but didn’t delve into the many decisions they were making at all, instead just assuming wrote, stereotypical depictions of medieval settings without established connections or thoughts about theme, pacing, tone, y’know the things that make a story a story.
They didn’t want art. They wanted a big data set of facts and assumptions, with all the biases that come with those assumptions. At least they were willing to pay for it.
The same is true with AI art programs. They don’t care about the art, or the artists. If they did, they wouldn’t rely on millions of images stolen from other people. They could have used public domain images. They elected not to.
Argument Two: AI Art isn’t Theft
Stealing the profits of my labour is theft. Welcome to the big wide world of Marxist Economic Theory! At my last job the owner of the pizza shop gave the workers a small percentage of the profits generated by selling the art (pizza). The pizza had no value until our labour was applied to the ingredients and tools and the restaurant to make it pizza. We transformed the consumption of that pizza into profit, but the profit was not shared equally. The owner hoarded that profit and arbitrarily decided how much of it the individual workers would receive.
I’ve seen it put forward by defenders of AI art that a word like theft obfuscates the discussion. I disagree. It’s the same problem. People take the labour of many more people and claim they have done something to earn the end result when they have done nothing more than obfuscate the process. The person who owned that pizza shop didn’t do anything transformative to create pizza, just like the people who type words into Midjourney don’t make art. They just steal someone else’s labour and call it theirs. Profiting off the labour of others is still theft.
Not to say that you are using these programs for profit. But it’s worth noting that people are, and the companies producing these programs absolutely are. In fact they are attracting venture capitalist investment from folks who think a great deal of money could be made with them.
Argument Three: My economic situation means I cannot afford to pay an Artist
My economic situation is the same, I barely afford to pay my bills every month, but I am generating enough profit from selling and Kickstarting games to pay those bills. How? By taking advantage of solutions artists have already created and ignoring art entirely:
1 Art Packs. There are affordable collections of art on DriveThruRPG and itch.io, and dedicated sites like Creative Market. Artists want people to use their work, and they want it to be affordable and transformable.
2 Photobashing. This is where you use Photoshop (or other free programs that do the same thing like Glimpse, or cheaper alternatives like Affinity) to mash public domain and creative commons photos and art from sites like Pixabay, Unsplash, etc. into the thing you want. It takes time and practice but that’s how I went from what awkwardly tracing a public domain image:
To photobashing several such images together into what I want with about 4 years of practice:
3 No art at all. If you're uncertain about whether to use AI art to subsidize your creative work I have great news for you: custom made or commissioned art is not required to communicate your vision in TTRPGs. In my experience there is very little correlation in terms of revenue or downloads. But I’m a writer, I prefer narrative and words over pictures so maybe that’s how my brain works. But in my time as a game designer I have found that expensive art is not required to get folks to download and read your game, not required for you to make money, and not required for you to get noticed or fund a Kickstarter.
Here’s the breakdown of my top viewed, downloaded, and sold games on itch.io, with all the games where there was commissioned art highlighted. Everything else either:
- Doesn’t have any art
- Has art I learned to make myself
- Has art from art packs I bought from artists
(Please note that External Containment Bureau wasn't my project or made just by me. I was one of the writers and designers along with Eric Brunsell, Eli Kurtz, Justin Ford, and art by Julianne Griepp)
Neon Black is one of my most popular games. It has over 60k words and 1 piece of art that was given to me by a graphic designer in exchange for letting them use the text of my game to do some layout work to bolster their portfolio. If you think you need art to compete, look at these images and ask yourself why the games with custom art are being outperformed so consistently across the board.
AI art isn’t democratizing anything. As long as you have a computer, digital art is already democratized. The tools are free, the materials are free, all it costs you is time. Time you get to spend learning and improving a new skill, meeting other artists and creators, getting advice, networking, and building a community. And even if it did empower people, it’s stealing labour from artists. And even if it wasn’t, even if it was a legitimate intelligence making art just like we do it wouldn’t fucking matter. Because it’s made by a bunch of rich people, backed by more rich people, who want to use you to get richer.
They don’t give a fuck about the art. Why do you?
