Your social media feeds have probably been flooded of late with bizarre yet detailed images you can’t quite explain. Are you seeing an Astroboy-style manga moment featuring Spider-Man at Woodstock? An oil painting wherein Harry Potter plunders as a pirate despite a suspicious number of fingers? Characters from The Office in a Mad Max world?
Chances are, if you’ve suddenly started noting such feverish weirdness, you’re probably looking at a piece of artificially intelligent art, or AI art. And though the tech and its more widespread use are both relatively new, at least at such a level, the implications of such new possibilities are dizzying.
But so much of what is and will happen is still unknowable. AI art has reached ubiquity on the internet, in media and in advertising so quickly, in fact, that no one is quite sure what to make of it, how it will continue to affect arts and artistic types, what legal ramifications might present themselves or even who owns what. SFR spoke with a number of people working in the arts, as well as an attorney, to try and understand the human response to AI art through the lenses of those who make a living through creation. The questions, though, might still be too muddy to answer in a satisfying way.
Nevertheless—the answers we heard are as manifold as the bizarre, often beautiful images that spawn them.
AI art in this context refers to images generated through deep learning artificial intelligence programs such as Stable Diffusion and other, similar algorithms that have been trained through a large and growing dataset. These programs constantly “learn” how to produce new images over time based on the images and text prompts users feed them. Many such apps have popped up and likely will continue to do so, but Stable Diffusion remains the biggest and most robust among them, for now.
The process is fairly simple: Feed an app’s algorithm your own images, or type out a description, and watch as they’re rolled up into randomly generated art pieces from billions of potential amalgamations. Over the last several months, AI art has laced our online lives with surreal worlds featuring hyper-detailed and sometimes gorgeous elements; perhaps stealthily, though, the tech has started producing works unrecognizable from handmade art—AI has even managed to believably imitate real photography.
It’s also imperfect. Discerning viewers who glance at the hands and limbs of the often otherwise beautifully rendered figures may sometimes note discrepancies in the number of fingers; the algorithms have struggled to produce believable faces, too. Someone who knows where and how to look can easily find red flags, and though such oddball moments have inspired no shortage of memes and internet jokes, apps including NightCafe, DALL-E2 and Deep Dream Generator have been learning at an almost alarming rate, and the so-called artworks have improved rapidly.
And why wouldn’t the process improve? Such apps are available to all, and users can do as little as enter a text prompt. They can go further, too, inputting a string of keywords, descriptions and specific art styles. This begs questions, though: Can a user with no discernible artistic skills be classified as an artist when the breadth of their effort was typing a few words into an app? Who decides what is legitimate art, anyway?
At nearly every point in human history, there are those who have resisted new advancements, new tech, new opportunities. Digital art created with PhotoShop, for example, was once widely regarded as lesser, yet now stands beside traditional artistry as its own accepted format. Television was going to rot our brains; the radio was going to corrupt us; the Gutenberg Press was going to put devil words in the hands of the icky proletariat. But humankind has always harnessed tech and bent it to our will. In this case, though, when it comes to the clash between arts, humanity and computers, something feels different. Something feels like it’s happening too quickly.
Nikesha Breeze, an interdisciplinary artist, activist and co-founder of Santa Fe’s Earthseed Black Arts Alliance, has a nuanced take.
“I think that [AI art] is going to have a lot of negative impact and a lot of positive impact,” they tell SFR in a phone interview from Ghana. “I think that just like any tool that comes in, there’s going to be a massive shift in how we think about art, how we think about media in general, how we think about image and accessibility and community.”
Breeze has proven adept at art forms from illustration and portraiture to large-scale sculpture and fabrication and says there’s a kind of allure in the idea of using AI generators as a launchpad, and/or in automating tasks to build off those images that use their existing handmade skills and artistry. Creating quick backgrounds or workshopping an idea without having to put in hours of effort sounds enticing, as does popping out a dozen potential versions of a piece in minutes. Still, Breeze says, they are optimistic but cautious about the ways they might incorporate the process into their own work—not to mention the possible implications for other working artists.
“I think it will continue to help me support and refine my art, to be able to use it as a tool to stretch the capacities and potentialities, speed, production—all of these things in art,” they say. “It…may be a useful tool I’ll implement in the ways any futurist artist would try to do. I’m curious and excited to see how I’ll use it and how I’ll continue to refine my own skill through it.”
Of course, that’s part of the rub. AI-generated art is still in its infancy, and not all artists see its rise as a beacon of progress—or even as an acceptable tool. Some see a slew of ethical issues, ranging from art theft to fewer jobs in a world that already has scant options for artists.
In a recent Editor & Publisher piece, for example, cartoonist Rob Tornoe lamented that, “The idea of a computer now creating a work of art that might take me several hours to produce feels like a punch in the face.”
Frank Ragano and Mariannah Amster of the annual CURRENTS New Media Festival view AI art’s possible effect on creative jobs as somewhat murky, and they should know. Together, they’ve created one of the country’s preeminent digital arts events in Santa Fe, and much of the success hinges on how artists use new and emerging tech.
“There’s no doubt it’ll change a lot of creative jobs,” Ragano notes. “I mean, people will either lose jobs or adapt to using AI. If a business finds that it can produce work for, say, advertisement, that is successful, and it costs them less than using an artist, they’re definitely gonna do that.”
Amster points out another possible effect.
“That may be true, but I think one of the things we’ve learned doing CURRENTS is that people still at first are really excited by the technology,” she explains, “but then the human hand or the artist’s hand comes back into the work and people want that physicality.”
At CURRENTS 826, the festival’s permanent Canyon Road outpost, for example, the recently closed Earth’s Other group show featured artists David Stout and Colin Ives using AI methodologies in print work—while maintaining their own notable and perceivable human touch.
“For me, and the artists we are seeing and choosing to show, it’s not the AI on its own making work. That would concern me. I see it as a tool for artists, or a collaborating partner,” Amster says. “I feel like artists are really coming into their own when it comes to using technology…it’s not just fascination with the technology, but really bringing an artful perspective to the technology.”
“The best results have been through AI and human artist collaboration,” Ragano adds.
Other creative types are less enthused.
Self-described “engineerish artist” Justin Michael Crouch, whose work has included large metal sculptures, has concerns. He and others have regularly spoken out against AI art on social media, and in real life, and he says he’ll continue to consider generative learning apps as problematic.
“Is AI ever going to give us another Da Vinci? Is AI ever going to give us another Michelangelo?” he asks. “Absolutely not, because all it is doing is looking at every single piece of art we’ve created and remixing it into something we’ve technically never seen—yet you can look at it and see the derivatives from real effort that is put forth by an actual consciousness.”
Here, Crouch touches on the sources of training for AI image generators. Stable Diffusion, for example, is built from a database of over 5 billion image and text pairings from the nonprofit open network, LAION. LAION’s stores are massive, and with users training the system daily, the collection will only grow. The universe of potential legal issues is staggering to consider.
Do all users obtain consent from the artists whose works they upload? Of course not, and many believe creating AI images—or, as Crouch says, remixing them—ultimately violates the intellectual property rights of the artists whose work upon which it was trained.
“As for the ethics of AI illustration, I find it alarming that some artists have had their art pre-empted without permission, license, royalty or commission to form the basis of some scammer’s AI illustration,” says Santa Fe-based collage artist Deco, who speaks from 40 years of experience in the arts.
Crouch, meanwhile, doesn’t mince words.
“When you plagiarize 2.4 million different things and then just rearrange all that plagiarism into something that seems original, it’s absolutely not because you can always go back and analyze the thousand different conglomerations that it has utilized,” he says.
Some are trying, however, including website DeLouvre, which helps artists seek out the possible inclusion of their work simply by uploading AI images they believe have done so. Even so, having to search out credit that way seems a challenge, and Crouch, Deco and Breeze still have fears.
“I have every ethical concern about AI art,” Breeze says. “I think an entirely new way of thinking about copyright, thinking about open source, all of these things, is coming into question. Ethically, there are major challenges, and I think that as the tool is growing and as the tool is being used equally, we need to begin to look at ways to support and protect ourselves and each other through its use.”
That brings up another minefield of considerations. Can an image not technically created by a human person be copyrighted or even just owned by a human person?
Way back in 2011, for example, wildlife photographer David Slater found himself embroiled in a legal quagmire after a macaque in Indonesia accidentally shot a selfie with one of Slater’s cameras.
“The smiling monkey that took a selfie has already laid down a precedent,” Crouch says. “The photographer who had that image on his phone wasn’t allowed to copyright it because a piece of art by law requires human authorship. Nobody can tie their ownership to it because it is not humanly authored.”
Eventually, though, Slater worked out how to monetize the image, at which point People for the Ethical Treatment of Animals filed a suit on the monkey’s behalf. Slater ultimately settled in the 9th US Circuit Court of Appeals and can make money from the image—he must also donate 25% of any revenue to groups that protect the macaque and its habitat in Indonesia.
Still—who owns what when we’re talking about artificial intelligence?
“It depends, is the answer,” says Talia Kosh, an attorney with Santa Fe’s Bennett Law Group and Albuquerque’s Blackgarden Law who specializes in intellectual property. “If a user enters, ‘Civil War landscape’ [into an app], the ideas themselves aren’t copyrightable—there can be variations of Civil War landscapes. Are they substantially similar to other works?”
Kosh says it’s even harder to nail down AI art law when it comes to fair use—a tricky corner of the law that allows for the use of copyrighted material without permission for things like criticism or parody.
“If I’m doing like the artist Richard Prince did, taking somebody’s original work and putting my own commentary on it, maybe making a joke or a parody, that can be considered transformative use,” she says. “There has also been case law that says two photographers can take the same photo of the same landscape in the same place and at the same time of day, but it’s about the angle, the lighting, the choices in composition—those are elements of originality.”
Prince printed out large scale versions of random Instagram users’ photos and sold them in galleries for absurd amounts in and around 2015. At the time, critics raised questions about ownership, and now even more than then, there’s no easy answer.
“[AI art] is so new that the law is still catching up,” Kosh tells SFR. “As suits are brought, the law will develop. The courts are much quicker to respond than the government.”
Santa Fe oil painter Amanda Banker, who has dabbled in NFTs, emphasizes the role of existing artists. Yes, she says, anyone can use AI, but those who have trained or self-taught or even just have more artistic experience will surely create more meaningful work than someone entering prompts they think are funny into an algorithm.
“What I’m interested in seeing…is the artists—the real artists—getting ahold of it, because I think that that’s where the distinction is gonna happen,” she says. “If I know artists, they’re gonna take it and they’re gonna run with it, and things that we have not seen yet, really incredible things, are gonna start to come out.”
In terms of creative jobs and the potential AI impact, Banker notes, “I think the people that actually have good taste will keep an artist around. The people who want to have a cheap, easy product, they’re the ones who are gonna drop the artist.”
The question remains for many artists though, whether AI or not: What defines “art” in the first place? If a person puts a prompt into an AI generator and it produces an image, is this person an artist?
Deco, the collage artist, has doubts.
“AI can generate illustrations, which may be beautiful, but cannot, by its definition, create to reflect the depth of the human soul,” she explains. “My art process is slow and painstaking and is entirely formed by hand. I am not afraid of technology; I do use a giclée process to reproduce my originals; I find this service expensive, but I pay for it willingly because it allows me to generate income. These giclée prints are beautiful and faithful to the original, but they are not ‘the real thing,’ not art.”
Banker takes the concept even further.
“What we think is going to happen is that it’s almost gonna be a Blade Runner situation, where physical art is going to become more and more valuable because AI artists are going to flood the market, and I think physical artists are actually going to end up being more valued,” she says. “I can go online and sign up for Stable Diffusion right now and generate however many images I ever want, could ever think of, and then print them to aluminum board—and they’ll look good, but it’s kind of like buying a poster at Cost Plus. These are millions and millions of prints made, so there’s really no individual value.”
Even if we consider human touch a necessary aspect of creation, Breeze points out how accessibility can be problematic, too. Who can afford access to the materials to regularly participate in physically created art?
“As a person and as an artist, AI is right now absolutely on the table,” Breeze says. “I’ve been talking about it here in rural Ghana working on a huge sculptural project in a foundry using these old techniques, and we’re talking about AI art, me and the other artists here—artists who work with their hands and traditional methods. I think that when we harness it, especially under-represented artists, artists of color, artists who don’t have a lot of accessibility to all of the tools of the world, when we are able to utilize some of the capacities of AI art, it’s basically a new language. It’s something we can harness.”
And, it seems, artists might have to learn how to do that. When it comes to AI art and generative apps, Pandora’s box has been opened—and there’s no shutting it now.
“I think it’s just a creative tool and I see it very similarly to photography, the computer, the resistance that people have against new technology. I think the power is the issue. Human beings tend to like to surrender to power,” Amster says. “So people have to stay on top of it if they care, and not surrender to the power of AI.”
And so it will go, in flux, at least for a time. Stable Diffusion is facing a lawsuit for using copyrighted artwork, and you’ll surely find many others in the works already; tech-savvy folks are already building new formats, too, such as ReachAI, which has an opt-out option for artists to request their work be removed from use in AI training. .
“We’ll just have to wait and see what happens,” Banker says. “It’s gonna be interesting.”