This is an essay responding to a post on social media, and it started just as a “SOMEONE ON THE INTERNET IS WRONG” essay, but it’s actually come around to a call to action. Here’s the tl;dr, if you read nothing else about it: AI art is a new tool with a lot of potential for both use and abuse. It’s threatening the livelihood of some furry creators. What you can do to help the furry art world and the furry community in general is: keep supporting furry artists. Keep getting commissions and pledging to their Patreon and so on. That’s what we can do to effect change in our little part of the world.
Okay. If you want the long version, here it is, starting with the post that inspired this essay.
“After watching the haters, I have a new theory:
“Artists who are good enough have nothing to fear from AI.
“Therefore, artists who ARE threatened by AI need to pick up a pencil and improve their skills.
Telling others to learn art skills is a projection of their own insecurities.”
The above was written by someone who has been part of the furry community for a while now, over a decade. If you’re like me, those statements don’t sit well with you. I’ve identified several areas that make me really uncomfortable with them, and the more I wrote them out in my head, the more tangled I realized they were. They intersect with a lot of other subjects, which I will make note of so you know I’m not unaware of them, but the purpose of this essay is not to litigate the ethics of how AI is sourced, nor the question of whether making art with AI constitutes “artistic expression” in the same way as picking up a pencil or an iPad with ProCreate. The purpose of this essay is to address the above statement, that if you’re threatened by AI, the answer is to improve your own skills rather than try to curtail the use of AI, with specific reference to the furry community (which I have also been a part of for a while). And as you can see, this introduction is already getting too long to hold people’s interest on social media, which is why I’m putting it here.
Before I dive in, a point of clarification. I have been and will continue to use “AI” to refer to the so-called “artificial intelligence” systems that use machine learning (ML) and training on an extensive dataset to produce creative output, usually at the direction of a human operator. In this specific case, I’ll mostly be talking about visual art, although the same argument can apply to prose or music or any other art (maybe not dance yet, but choreography?). I’ll use “OP” (original poster) to refer to the author of the above quotes.
Ready? Okay, let’s go.
1. How Empathy Works
Here’s the one that actually makes me angrier than any of the other parts, that makes me want to title this episode “FUCKING EMPATHY.” But I’m trying to be measured, so I’ve allowed myself the rhetorical device of “if I were being naughty, I would’ve cussed,” which yeah, is a cheat, but gives me at least a little bit of satisfaction.
First of all, “I think the people who disagree with me about the use of this technology just aren’t very talented” takes the argument from a discussion of ethics to a personal attack. It’s akin to James Comer saying, “Well, the reason these people keep asking me tough questions about the investigation I’m conducting that has no evidence and can’t even name the crime it’s investigating is because they’re pandering to a low-IQ audience.” It’s sidestepping the legitimate questions entirely to cast doubt on the questioners (whom OP has categorized as “haters,” a neat trick to dismiss their concerns, because “haters gonna hate,” what are you gonna do?).
Second, this is a thing tech bro AI art apologists have said a lot. For example:
The difference here is that when the tech bro AI art apologists say it, they’re talking to an anonymous crowd of artists out there who are raising concerns about what AI art might mean to them. But the tech bro AI art apologists aren’t friends with those artists. They don’t go to the same conventions, share the same publishers and culture.
OP has pulled out this “pick up a pencil” line to use in response to the furry community, which they are supposedly part of. One of the best parts of the furry community and the thing that’s kept me here for thirty-plus years is the way we all look out for each other. I have too many examples to list, and anyone who’s been in the fandom for any length of time will remember examples.
(The exceptions, as far as I recall, come when people harm others in the community—by tracing/copying art, by assaulting others, by refusing to pay for/complete commissions—and refuse to own up to the harm they’re doing.)
So even before we dig into whether this is well-reasoned advice, let’s just look at the context. The broader issue is that the use of AI is reducing the market for furry artists by providing people with a free alternative (the most high-profile example of this is a furry ‘zine whose editor uses AI art rather than license illustrations from furry artists). This means that furry artists stand to see their income reduced, and it’s not hard to envision some people who had previously been earning a living with their art becoming unable to continue doing that.
If that’s true, even if your advice to them is “Then just get better at art,” it doesn’t hurt to show a modicum of fucking empathy for what they’re going through.
Sorry.
If your friend’s house was in the path of a wildfire and they called you in a panic because they had to evacuate, would your response be, “I guess your old house wasn’t any good, time to get a new one”? Of course not. And yet, here we are.
(I don’t want to speculate too much on OP’s state of mind, but I know that after a day or two of intense arguing on social media with a bunch of people arrayed against me, not to mention a history of months of having such arguments, I might be in a place where empathy was hard to access. But. That’s when it’s most important. And if you really can’t access that empathy in a situation, maybe…just maybe…you don’t need to come out with a new theory.)
Look, it doesn’t take a college professor to connect the dots between the theory that “people who hate AI can’t produce art as good as AI” (which is also flawed, btw; I know plenty of people who can produce way better stuff than I’ve seen come from AI who are also not fans of the technology) and the realization that “oh, those people are scared of what AI means to their future.” And responding to scared people with “suck it up and get better” isn’t even tough love. It’s just tough.
2. How Art Works
The reason “pick up a pencil” has become such a fun line for tech bro AI art apologists to say is that it shifts the burden for coping with the problems AI art is causing onto the people most affected by it, rather than being shouldered by the people who are, you know, causing those problems.
Whenever a tech company throws an innovation out into the world and claims to be “disrupting” an existing marketplace, a lot of these same ideas get thrown around. “The old way is inefficient,” and “you have to break things to move forward” and all that kind of bullshit. People are now trying to apply that thinking to the AI art sphere, and so the implication behind “pick up a pencil” is that the real problem is the lazy, entitled artists, who have been stagnant for so long because there was nothing to challenge them, that people have been paying too much for average art and AI is now here to deliver them from this nightmare.
But that’s not the problem being addressed by AI art. That problem is, as is every problem addressed by tech companies, how to take money that is being spent in one place and transfer it over to the tech company. In this case, they have tried to undercut the market* for art by allowing people who haven’t spent years honing their skills to create pretty pictures.
(* Because I have nowhere else to put this: they can undercut the market because they have a lot of funding, but also because they have not been forced to pay licensing fees to all the artists whose work they used to train their datasets. If they did, their rates would probably not be competitive with furry artist market rates, at least. Whether this dataset training can be considered “fair use” is currently being considered by the courts.)
You see the difference? The problem isn’t that people have been paying for substandard art; the problem is that (in the eyes of the tech companies) people have been paying the wrong people for art. So how is “pick up a pencil” going to help an artist in that world?
(Besides that, the whole “just pick up a pencil” is a foolish thing to say when AI art has been improving by leaps and bounds. In just the last year it’s gone from “six-year-old trying to imitate art” to producing at least average quality stuff. In six months, it will be technically even better. It’s improving at a rate faster than most humans can improve their art.)
It also severely downplays the impact on artists. In a very privileged way, it assumes that all artists just have unlimited time to practice and improve. In this capitalist society we live in, though, artists have to pay for their own education. In society at large, artists can get jobs to improve their craft once they reach a certain skill level, but for an artist to support themself making their own art…that’s rarefied air.
Except in the furry fandom.
In the furry fandom, the demand for art is such that we are funding the development and education of a relatively huge number of artists. It’s an art-forward fandom and there are a ton of artists who make a living off of commissions and other art projects.
So when a new development comes along that threatens to take away the market for some furry art, it is also reducing the number of people who can devote full days to improving their art. If someone can’t earn a living from their art the way they used to, they can’t just “pick up a pencil” and improve until they get “good enough” to earn a living again. They have to eat and pay rent and medical bills in the meantime. They can maybe still work on their art, but more slowly, and besides, there’s a rush of joy that comes when you’re making art you love for people who love it.
And that means that the number of people working on improving furry art as a whole is going to decline. Who gets to keep improving their art? You already know the answer: people with privilege.
I don’t know OP’s circumstances, but “pick up a pencil” is a thing that comes from a place of privilege. It assumes that everyone has the means to pursue the thing they love. Not everyone has a family that can subsidize their art. And art is already biased toward those with privilege. I came from a stable middle-class family that was able to send me to a good school. I never had to worry about where my next meal was coming from or whether it was safe to run to the store. I could sit oblivious to the world in my room and read. And I could go to a good college, where I could earn a living at a job that afforded me enough free time to work on my writing in the evenings, take workshops and classes. I have a family that supports me. I’m extremely privileged here.
So that’s the thing about saying “sure, AI is going to replace some artists in the fandom, but they can just improve their skill until they’re ‘better.’” Not everyone can just go back to the drawing board. We’re going to lose some people, and the people we’re going to lose will be the ones without privilege and money—in other words, the people whose art is most different from the majority being created already, the people whose art is already difficult to find and access.
3. How Tech Works
AI’s development is being paid for by large tech companies. If you’re not familiar with how tech companies try to drive adoption of their product, here’s a quick refresher: they offer it for free or at a discount at first, to undercut the existing market that has been in place (see: Uber most famously, but also nearly everything). Then, when they’ve built a large customer base for their product, they raise the cost, either with microtransactions, by raising the price directly, or by selling advertising or by selling your data (see: Facebook, Twitter). AI art is free to use right now (but not to make; it’s estimated that every 1000 AI images generated is the equivalent of driving a gas-consuming car four miles, which doesn’t sound like a lot until you think about how many iterations it takes to produce a usable image and how many people are doing it), but do you think these for-profit companies are offering it for free because they want to contribute to the public good?
Already, over on Bing’s image generator, you can use 15 “boosts” per day to accelerate the creation of the images you create. If you want more, you can “redeem Microsoft Rewards” for more boosts. As these companies look to recoup their investment (the charitable read) or just boost their profits (probably the more accurate read), tell me if any of these possible future developments sound crazy or unfamiliar:
· You can create a few images a day for free, but if you have a Premium subscription you can get ten times as many!
· Our ArtBro 3 engine is free to use, but the much better ArtBro 4 engine is available only to our Premium subscribers.
· With a Premium subscription, our TalkBro 4 AI engine will help you create a better prompt for our ArtBro 4 engine, giving you images closer to what you’re looking for in less time!
This is the future of AI art. And yeah, none of those things have necessarily happened yet (psych! ChatGPT4, much better than ChatGPT3, is available…for a monthly subscription fee), but I’m not exactly going out on a limb predicting this. I’m basing this on tons of examples in recent history, from Uber to mobile games to web browsers. Companies are able to offer their AI art engine for free because they have hundreds of millions of dollars in funding, and they have that funding because people are betting that they will be able to make that back. Sooner or later, they’re going to be asking you for money.
4. How AI Works
Here’s the other thing about AI: it produces its images based on a dataset. It can create an image that has never existed before, but it’s creating that image out of the same building blocks every time. And those new images maybe go back into the dataset to train the AI on elements it’s already using, but they don’t offer it anything new. It just becomes more inbred, refined versions of the same thing.
You know what does create something new? Human-drawn art. However ethical it is for companies to use art to train their engines on without compensating the artists (I think you can guess from the last section that I think it’s pretty unethical), they can’t add new artistic elements to those datasets without more human-drawn art. So the more the AI bros succeed in driving people out of the art business, the more they are stunting the growth of the product they love.
By contrast, if you pay for furry art, you are not only paying for a piece of original art that can incorporate new elements, you’re also paying that artist to improve their skills and keep producing art for our community. The more we pay furry artists, the better our furry art community grows and the more art is available for all of us. That one artist with a funky style who’s barely scraping by can afford to keep practicing and refining their style until it becomes something amazing. Their work, in turn, inspires other artists to grow and develop their own style. Every dollar or Euro or pound or whatever you spend on furry art multiplies in value to the community.
AI art can be produced “in the style of” any popular artist. What it can’t do is come up with its own style.
The Conclusion
Back to wrapping this all up. What does all the above mean? Being part of the furry community gives us a unique opportunity to affect our own future. We can’t stop the growth of AI art and most of us have precious little input on whether marketing firms/studios/design firms use AI art, even if we feel strongly about it. The most we can do is write letters to the people deciding these things (and do write to them when they come up, like when the U.S. Copyright Office was holding hearings about whether to allow AI art to be copyrightable), which does move the needle, but in a very small way, largely imperceptibly and unsatisfyingly.
What we can do, and what actually makes a huge difference, is keep paying furry artists and supporting the furry art community. Get commissions, subscribe to Patreons or Ko-Fis, buy their merch. Every bit of support goes back into our community and helps those artists improve their art and adds more art to our community and culture, both now and in the future. You don’t have to argue with OP or the other AI art apologists, you just have to throw your support behind our furry artists, and encourage friends to do the same. Furries support furries—it’s what we do. And I believe we will keep our artists going, because we don’t just like art; we like seeing how other people interpret our characters and ideas. And we care about our artists, because they’re our friends.
We can create the world we want to see, and we don’t do it by using a program some big tech company created. We do it by supporting our community.
do you like doing art or digital?
The linked article is half a year old and is more of an opinion piece than a reflection of reality.
Synthethic datasets often perform better than the raw stuff, as exampled here https://arxiv.org/pdf/2401.00368.pdf as well as here https://arxiv.org/pdf/2401.01335.pdf , as well as community driven projects in image generation.
ML can also be used to discover new things, like here: (Mathematical discoveries from program search with large language models) https://www.nature.com/articles/s41586-023-06924-6 or combine old concepts to create new ones.
I'm not going to bother arguing with the rest of your article, but I think everyone agrees that it is better to rely on facts rather than myths.