What Happens When AI Art Starts Selling?

June 26th
(Obvious, an AI art group)

This post is sponsored by Brilliant, which offers courses like Cryptocurrency, Computer Science Fundamentals, Logic, and The Joy of Problem Solving. Sign up for Brilliant through Diagram and get 20% off today

Occasionally, the same people who warn about automation killing jobs also try to offer some solace — humans, they say, will have more time to be creative in the future, once the menial jobs are all taken by robots.

That may not be true. People may have some competition in the arts, too.

AI has been used to create news stories, music and poetry. It’s also the tool creators and trolls use to create deep fakes. And, more recently, it’s been utilized by artists to create “paintings” that, in one case, have sold for $432,500.

Every time AI enters an industry, it sets off questions about human ownership, invention and creativity. In gaming, AIs have been required to only receive information equivalent to the speed of a game’s framerate — about 30 screens per second — to make the information it receives equivalent to that of its human opponents. Theoretically, a gaming bot could take in much more information than that.

It’s happened in patents as well. Patent applications for AI-developed inventions have hit roadblocks in the US and Europe, which had no precedent with which to handle an invention with no direct human owner.

Many institutions and industries don’t know what to do when working with this new technology.

Much more so with art.

“Everybody has their own definition of a work of art,” said Gauthier Vernier, the founder of an AI art collective called Obvious. says.

In an interview with Time, Gauthier defended his organization, which trains AIs on art pieces and then generates composite pieces.

“I’ve tended to think human authorship was quite important—that link with someone on the other side. But you could also say art is in the eye of the beholder. If people find it emotionally charged and inspiring then it is. If it waddles and it quacks, it’s a duck.”

To date, Obvious has sold several pieces, including a trippy portrait that resembles a portrait and sold for $432,500.

To create their art, Obvious collected 15,000 portraits from the 14th to the 19th century and trained an AI on them. As it created composite work of its own, Obvious’s “Discriminators” judged the originals and tweaked the algorithm.

““Think of it as the 15,001st image,” then-PhD student Hugo Caselles-Dupré told Time. Caselles-Dupré built the algorithm.

In other cases, AIs work with the artist. Suogwen Chung, an illustrator, has produced dozens of pieces alongside a robotic arm that reacts and draws alongside her in real time.

“The robot mimics the artist like a partner in improvisational round singing performance,” Chung says on her site.

Others still see AI as an instrument in the traditional sense of the word, like a paint brush or camera. Mario Klingemann, who also sold his AI art, said that his tools -- neural networks, code -- aren’t any different than another artist’s equipment.

“If you hear somebody playing the piano, would you ever ask, ‘Is the piano the artist?’” Klingemann asked in an interview. “No. It’s the same thing here.”

Klingemann calls his most famous work, Memories of Passerby I, a “machine.” It generates changing imagery on two screens embedded in picture frames. Klingemann says that the art is not the imagery itself. It’s the code.

Klingemann coined a term for his art’s style -- the dreamlike, surreal look of his figures, which seem to morph into one another. He called it the “Francis Bacon effect,” and said it applies to a lot of AI-generated art, be it from Obvious or Google Deep Dream. The effect comes from their use of the same GAN algorithm.

Like Obvious’s work, Klingemann’s Memories of Passerby I sold. It was listed by Sothebys and was purchased for £40,000 in 2019.

But even with this early success, critics already expect that the Francis Bacon effect will haunt the movement, which Obvious calls “GAN-ism.” When art can be produced immediately, from just a click, what happens to the scarcity on which traditional art has partly relied?

“They’ve started to think about this stuff beyond the ‘selling point’ of it being art produced by AI, but have also started to think about how conceptually rich can they make it,” said Karthik Kalyanaraman, an economist and art collector who created the first ever AI gallery show, Gradient Descent.

Kalyanaraman’s show led to multiple purchases and received an especially enthusiastic response in New Delhi, India.

“The more we fear something, the more we get controlled by it,” said Aparajita Jain, the director of the Delhi gallery who hosted Gradient Descent. “If we understand it, we can use it as a tool.”