He used AI to win a fine-arts competition. Was it cheating?

by Drew Harwell / The Washington Post

15:52 JST, September 3, 2022

When Jason Allen submitted his “Théâtre D’opéra Spatial” into the Colorado State Fair’s fine-arts competition last week, the sumptuous print was an immediate hit, beating 20 other artists in the “digitally manipulated photography” category to win the first-place blue ribbon and a $300 prize.

What Allen had only hinted at, however, was that the artwork had been created in large part by an artificial-intelligence tool, Midjourney, that can generate realistic images at a user’s command. The portrait of three figures, dressed in flowing robes, staring out to a bright beyond, was so finely detailed the judges couldn’t tell.

Allen’s piece offers a clear example of how rapidly AI-generated art has advanced. Trained on billions of internet images, the systems have decisively pushed the boundaries of what computers can create.

But it has also sparked a massive debate over the meaning of art, with Allen facing accusations that he had been deceptive in beating out other human artists with something he asked a machine to create.

Text-to-image tools like DALL-E 2 and Midjourney have quickly increased in sophistication and become one of the hottest topics in AI. They can not only just generate fake people, objects and locations but mimic entire visual styles. Users can demand the art piece look like a cartoon storybook or a historical diagram or an Associated Press photograph, and the system will do its best to oblige.

But AI-generated art has been criticized as automated plagiarism, because it relies on millions of ingested art pieces that are then parroted en masse. It has also fueled deeper fears: of decimating people’s creative work, blurring the boundaries of reality or smothering human art.

Allen said his art piece shows people need to “get past their denial and fear” of a technology that could empower new inventions and reshape our world. The AI, he said, “is a tool, just like the paintbrush is a tool. Without the person, there is no creative force.”

But he also struck a note of defiance at the critics he believed did not appreciate the profound message in his AI-generated art.

“You said AI would never be as good as you, that AI would never do the work you do, and I said: ‘Oh really? How about this? I won’,” he said. “It’s here now. Recognize it. Stop denying the reality. AI isn’t going away.”

***

Allen, 39, lives south of his hometown of Colorado Springs, Colo., and runs a company, Incarnate Games, that makes tabletop fantasy games. He went into the Air Force after high school and got a computer science degree at a Colorado technical school. He does not consider himself an artist and had never entered an arts competition before.

Earlier this year, he said, he started noticing people posting more AI art on social media, but he had been initially skeptical to try it himself, for “spiritual reasons.” Elon Musk, he remembered, had compared AI to “summoning the demon,” and the practice felt like it could be “a gateway into communicating with the unknown.”

“That’s what it feels like. This isn’t being created by a human,” Allen said Thursday while watching a jujitsu tournament in Las Vegas. “There’s a lot to be said about the spiritual meddling of our reality. They would use anything they could to influence humanity.”

But the art was just so intricate, he said, that he couldn’t stop thinking about it. He started playing with AI-powered art tools: Wombo Dream, NightCafe, Starryai. Then, someone invited him to Midjourney, and he became obsessed.

Midjourney has become one of the most popular AI art generators largely because it allows anyone to freely create new images on command. Using the prompt “/imagine,” users can type in whatever they want to see and the AI will return four newly created images in 60 seconds. The user can also ask the AI to improve, or “upscale,” the visual quality with new variations on the same idea.

The start-up, which calls itself “an independent research lab . . . expanding the imaginative powers of the human species,” operates largely out of a 1 million-follower network on the chat service Discord, with rooms devoted to character creation, environments and “show and tell.”

After paying for a corporate account, Allen started generating thousands of images, changing the text prompts with every creation. He experimented with new settings, scenarios and effects. He asked for images in the styles of Leonardo da Vinci and the American psychedelic artist Alex Grey.

The pieces that really caught his attention, though, were what he now calls his “space opera theater” series. He started with a simple mental image – “a woman in a Victorian frilly dress, wearing a space helmet” – and kept fine-tuning the prompts, “using tests to really make an epic scene, like out of a dream.” He said he spent 80 hours making more than 900 iterations of the art, adding words like “opulent” and “lavish” to fine-tune its tone and feel. He declined to share the full series of words he used to create his art, saying that it is his artistic product and that he intends to publish it later. “If there’s one thing you can take ownership of, it’s your prompt,” he said.

“I was like: Dude,” he said. “This is so sick! I want to see more of it! I’m addicted! I’m obsessed!”

When he found images he really liked, he pulled them into Adobe Photoshop to remove visual artifacts. In one image, the central figure was missing a head, so he also painted in a crop of dark, wavy hair. He used another machine-learning tool, Gigapixel AI, to increase the photos’ quality and sharpness, then printed the three pieces on canvas – all variations on the French phrase for “space opera theater,” which he thought sounded cool – and drove to submit them to the state fair.

When he looked at the pieces, he said, he saw “a supernatural reality . . . something we haven’t even been able to experience yet, past the great beyond.” But the pace of AI art is moving perhaps even faster than the internet. “You’re looking at art from a month ago,” he added. “In technology terms, that’s decades. This piece is antiquated compared to what Midjourney is doing now.”

***

The state fair in Pueblo, Colo., was an unlikely place for the writing of a new chapter in art history. The 150-year-old festival, known for its horse and livestock competitions, runs a series of more-traditional art competitions, including for homemade dolls, quilts, porcelain art and needlework, as well as for the best canned carrots, medicinal remedies and holiday breads.

Of the 596 entries in the “fine arts” competition, 21 amateur “emerging artists” submitted pieces of “digitally manipulated photography,” one of the fair’s newest categories. Asked what art materials he had used, Allen told state fair officials only that he used Midjourney – though he did not exactly go into detail, and no one seems to have asked.

One of the judges, Dagny McKinley, an author and art historian who runs a playwright festival in nearby Steamboat Springs, remembers walking past Allen’s canvas and being immediately drawn to a piece that felt reminiscent of Renaissance art.

“It had an immediate story: People looking out into another world, everyone with their backs to you, no one facing or engaging with the viewers,” she said. “You get interested: What are they seeing?”

McKinley said she did not realize the art was AI-generated but said it wouldn’t have changed her judgment anyway; Allen, she said, “had a concept and a vision he brought to reality, and it’s really a beautiful piece.”

Sebastian Smee, The Washington Post’s art critic, said the piece’s textures and lighting are reminiscent of Gustave Moreau, a late-19th-century artist associated with the Decadents, who influenced Edgar Degas and Henri Matisse. (He also recalled a quote from the artist Sol LeWitt, who said, “The idea becomes a machine that makes the art.”)

When Allen announced his victory himself on Midjourney’s Discord channel, saying he had spent “many weeks of fine tuning and curating,” the responses spiraled between muted excitement to outright dread. In a chat board devoted to philosophical debates, one user compared the win to “entering a marathon and driving a Lamborghini to the finish line.” Another user wrote that the “stunt” threatened to “get this tool banned and hated even more.”

The win also triggered a flood of rage online. A tweet expressing profane dismay at Allen’s win has been liked more than 85,000 times. Another person tweeted, “We’re watching the death of artistry unfold right before our eyes.” Allen said he has also gotten “a lot of very nasty hate” in his online inboxes. On his game company’s Instagram profile, one user said he should return his award and “post a public apology before some MAJOR backlash comes your way.”

Some of the frustration has come from how the tools were built: a similar tool, Stable Diffusion, was “trained” on 2 billion images taken from the internet, including from personal blogs and amateur-art sites such as Flickr and DeviantArt.

Allen, however, dismisses the point as shallow: “Where did you learn how to do your art? You looked at art. Whose art was it? You learned their techniques, you studied their art, you added it to your repertoire,” he said.

Allen listed the pieces for $750 each, he said, and two were sold at the fair to unknown buyers, though he now frets that he should have charged far more, given that it could be “essentially a piece of art history.” On Discord, some users asked whether he should have been more explicit, to which Allen replied: “Did I have to?”

Despite the online furor, Allen’s neighbors seemed more sanguine about adapting to AI. As far as anyone at the Colorado Department of Agriculture can tell, Allen did not break any rules. Pieces for the category are required only to involve “technology as part of the creative or presentation process.” Digital filters, color-manipulation tools and the “recombination of images” are all expressly allowed.

No one has filed an official grievance over the result, either, department spokeswoman Olga Robak said, though there has been an unrelated dispute in the fair’s goat-shearing contest.

Robak, who studied art history, finds the controversy fascinating. “People put bananas on the wall and called it art,” she said. “Even photography was not considered an art form for a long time; people said it was just pushing a button, and now we realize it’s about composition, color, light. Who are we to say that AI is not the same way?”

***

Jessica Hair, a 25-year-old receptionist at a doctor’s office who won third place in the competition, said she did not feel Allen had acted unfairly and had no hard feelings about his win.

Hair said her “Judge, Jury, Executioner,” which depicts a tuxedoed skeleton on a golden throne surrounded by skulls, took 15 hours to create with a stylus on an iPad Pro. But Allen’s piece took time, effort and subjective judgment, too, and “how do we qualify what is and isn’t art?” she said.

She did wonder, though, whether it might have broken the fair’s rules requiring all art to have been made by Colorado residents. Would the AI, wherever it exists, qualify?

McKinley, the contest judge, said she understands the frustrations from some artists feeling spurned in their craft and believes the festival should consider a category purely devoted to AI art. But she sees such technology as opening up a new world of possibilities for artists – and as something it would be better to embrace, since it isn’t going away.

“It’s not going to take away from a beautiful painting or a sculpture you can touch,” she said. “It’s just one more tool we have to advance what we can create.”

Gregory Block, an oil painter in Denver who was not a part of the competition, said he finds it hard to imagine an AI generator supplanting the hundreds of hours – and all the “heart, soul, blood, sweat and tears” – he’s invested into his art. But he also thinks back to the artists that first inspired him, who used rudimentary devices like the camera lucida, in the 1800s, to aid their own designs.

“That was thought of as cheating, too,” he said. “Yet they used it to make these incredible paintings: anatomically correct figures, beautiful soft lighting. . . . Those steps in technology are elemental to our art. Otherwise, we’d still be doing cave paintings with just our hands and blood.”

The AI, he said, can imbue the art with a mysterious beauty, made all the more special because it is so hard to understand. But “the soul any of us can find in a piece of artwork, the emotion, the human struggle we identify with in art is always our own.”

“It doesn’t have to necessarily be created out of a human soul, the artwork itself. It is for us to see and react to,” Block said. “We the viewers are, in the end, the ultimate artists. We’re the ones creating the world that is coming in through our eyeballs. That world is in our mind.”