The user posted a tweet seeking help to “emulate” the look of such acclaimed movies as “The Secret of Kells” and “Song of the Sea,” writing in December: “Is there a trick I don’t know about?” Could such an imitative image be produced using AI?
Moore tweeted his pithy reply: “Or you could learn to draw.” The cutting remark stirred a flurry of responses, striking at the center of the roiling controversy over AI generators that allow users to enter text prompts to produce algorithm-driven images, sometimes to mimic specific artists’ styles. These software tools have sparked such questions as: Are they ethical? Are they legal? And just what defines a digital image as art in a pixelated age of ever-blurring lines?
“I think a lot of us who studied and worked for years to achieve our style and our signature process felt offended by the idea we would just become something like an Instagram filter for people to apply to their AI prompts,” Moore says via email. He was irritated because his studio’s hand-drawn animation — which lovingly nods to Art Nouveau and medieval art — was being reduced to a superficial “cheat code.”
Moore’s view is shared by many professional illustrators and other artists who think that the rise of AI images, particularly within the past year, poses a creative invasion.
“I feel like it’s a violation of the soul, to be honest,” says Sarah Andersen, the Oregon-based cartoonist (“Sarah’s Scribbles”) and best-selling author (“Fangs”). “My work, like every other artist, is informed by my lived experiences and my education and my life and therefore, it’s deeply personal.”
Andersen perceives AI art generation as a legal violation, as well. So much so that she became a plaintiff in a class-action lawsuit in the Northern District of California that could have ramifications across the world of AI art.
In January, Andersen and fellow artists Karla Ortiz and Kelly McKernan filed the suit against Stability AI, which helped develop the open-source AI model Stable Diffusion; Midjourney, which has used Stable Diffusion for its text-to-image generator; and the online creative community DeviantArt, which uses Stable Diffusion for its own image generator.
To become trained, AI art generators rely on billions of existing images that are gathered — or “scraped” — from across the internet. Ethical and legal issues are coming into play with unrestricted, open-source AI tools because many of the images are under copyright and are used without permission.
“AI has stolen my work,” Andersen told The Washington Post by phone, and “broken it down in a way that I believe it will be able to create perfect copies” eventually.
Once AI generators can produce indistinguishable copies of art by humans, they become even more of an existential threat to the livelihood of artists, say Andersen and other illustrators.
Stability AI, though, wrote in a blog post last year that Stable Diffusion “will allow both researchers and soon the public to run this [tool] under a range of conditions, democratizing image generation. We look forward to the open ecosystem that will emerge around this and further models to truly explore the boundaries of latent space.” (Stability AI, Midjourney and DeviantArt did not respond to The Post’s requests for interviews.)
Dave McKean, the British artist, graphic designer and rock-star comics illustrator (“The Sandman”), sees AI as no minor trend. Rather, he calls it “an almost evolutionary shift in our culture.”
Once McKean began to explore how AI-generated imagery worked and how it could rapidly produce unceasing galleries of end results, he says he “spent a day on the floor of my studio in a fetal position.”
“My immediate response was: Well, that’s my career over,” he says. “Why would anyone pay me to do an album cover when anyone can type a few words into Midjourney and, in a couple of minutes, start downloading endless finished possibilities for free?”
In his experience in the illustration industry, McKean — who last year authored a book “Prompt: Conversations With AI,” about artificial intelligence and the creative process — says power has shifted away from artists. To those decision-makers who value high speed over creative originality, “AI is their wet dream.”
Organizations that represent professional artists, however, are looking to dissuade use of AI images. The National Cartoonists Society and the Society of Illustrators have released statements to try to protect their members’ interests.
Jason Chatfield, president of the National Cartoonists Society, says that many artists are primarily looking for credit, consent and compensation from AI art generation companies. He adds: “Legislators will take forever to even fathom the tech, let alone build in any guardrails, so it takes things like litigation and public debate to move the needle on the ethical use of this technology.”
Tim O’Brien, the past president of the Society of Illustrators, warns against allowing unrestricted AI art generators that, for instance, permit artists’ names to be used as text prompts: “We can shrug or do the braver thing: To come out in favor of human creators and not accept this form of artistic automation.”
Arvind Narayanan, a Princeton professor of computer science, paints a similar picture of the conflict zone between AI and many creatives.
“The companies behind AI art generators have developed and deployed them in a way that is hostile to artists, such as by scraping training images without consent or compensation,” Narayanan says. “Allowing the tools to produce images in the style of a particular artist seems like a clear case of appropriating the labor and visual distinctness of an artist.”
In Narayanan’s view, though, it didn’t have to be this way.
“The developers could have treated artists as partners and stakeholders, rather than raw material to train on,” he says. “Those who claim it was inevitable are simply making excuses for the failure of companies to develop the tech responsibly.”
The Society of Publication Designers also released a statement last month, to stand “with our illustration colleagues.” The organization, many of whose members hire artists, says its views on AI are part of an ongoing conversation.
The SPD board says the group is “built on the relationship between the artist, the art and photo director and reporter, so we want emergent technology to add to those relationships, but not take away from the artist.”
Ideally, the board told The Post via email, “There are some guardrails put in place on where the AI [generators] can pull from, either self-imposed or through legal regulation.” That could mean creating “a bank of images and artworks that is a mix of public-domain images, user-generated content,” as well as a setup “where an artist has opted-in and collects on a per-query basis,” akin to television writer royalties. Andersen, who helped bring the Northern California lawsuit, says she can envision a similar system.
Some creators of visual humor are feeling safer than other illustrators. The question at hand: Can computer technology create and “tell” a joke? Tjeerd Royaards, an Amsterdam-based editorial artist and editor in chief of the comics site Cartoon Movement, says the several dozen political cartoonists his site recently queried weren’t especially concerned about AI. “I think most cartoonists don’t feel threatened (yet), because AI is currently not able to produce satire,” Royaards wrote last month on his site. He noted: “I expect there is some way to go before AI is able to match a skill and wit of a good cartoonist.”
Aside from protection, credit and compensation, some artists say they are concerned about whether AI tools are capable of producing images of creative merit.
Painter and cartoonist Carson Grubaugh touts AI as “the largest development in art since the first hominid spat pigment at a wall to outline its hand.” He also disagrees with those who say AI-generated imagery is not “art.”
Grubaugh created what he calls the first comic book series in which the art is generated solely by AI, collected into “The Abolition of Man: The Deluxe Edition” due out later this year. Grubaugh, working with publisher Sean Michael Robinson, used the tools to interpret writings by C.S. Lewis and Luciano Floridi.
“The only way forward for artists and educators,” Grubaugh says via email, “is to totally re-evaluate how to foreground the emotional and psychological benefits of struggling to get good at something in a world where the end result can be reached with expediency.”
Moore, the Irish animator, does not view AI art generation as an existential threat to all artists, but does foresee that it will affect many aspects of visual publishing, including budgets and deadlines. He has played with the technology to learn about his “enemy” and finds it to be “honestly flabbergasting” how “something I had always felt was very human and complex can be automated so well.”
Moore is also still sizing up what AI generators are: “I sometimes wonder if it’s even less a tool and in fact a low-cost worker. Is it a robot artist or a very advanced paintbrush? I feel in many ways it’s a robot artist.”
McKean, the British artist, says that for him, creating art is about the experience traveled more than the finished work.
“It doesn’t vomit up art by any definition of the word that means anything to me,” McKean says of AI text-to-image software. “Art, for me, is a process — it’s not just about the end result. Making something involves intent, context, story, journey, testing yourself, growing. … The arrival is the least important part of the experience.”
So although AI can produce a “technically astonishing tsunami” of images, McKean says, he began to dive into the text-to-image generation and soon felt a distressing sensation — that in this space, individual human creativity is meaningless.
“If everyone can draw like Michelangelo,” he says, “there is no Michelangelo.”