AI-Generated Art Sells for Nearly Half a Million Dollars at Christie’s

It's the first auction house sale of art created by an algorithm

'Portrait of Edmond Belamy' sold for nearly half a million dollars.
Headshot of Patrick Kulp

A new painter with a very methodical style just made a big splash in the art scene.

Famed British auction house Christie’s just sold a portrait generated by an algorithm in the first-ever transaction involving an art piece created by an automated entity. The somewhat blurry but discernible painting, Portrait of Edmond Belamy, netted $432,000, beating the auction house’s expected price of $7,000 to $10,000 many times over.

The art was created by a French collective of artists and artificial intelligence researchers called Obvious with code mostly borrowed from a 19-year-old AI artist and programmer named Robbie Barrat. The group fed a system called a Generative Adversarial Network with a data set of 15,000 pieces painted between the 14th and 20th centuries. One component of the network would then paint its own work based on what it saw and try to fool another algorithm, which was programmed to tell the difference between real and algorithmically generated art. Once the latter could no longer tell the difference, the work was deemed finished.

The subject of the painting, Edmond Belamy, is a member of a fictional family of 11 of the collective’s machine-created portraits that resemble the type of aristocrats you might see in classical European portraiture. While the complexities of facial features typically make portraits tougher for computers to mimic than other forms, the collective felt it was the best way to show off their algorithm’s proficiency, according to Hugo Caselles-DuprĂ© of Obvious.

“We found that portraits provided the best way to illustrate our point, which is that algorithms are able to emulate creativity,’” Caselles-DuprĂ© told Christie’s.

While Obvious may have been first to complete an auction sale of AI art, they are far from the only creators attempting to use AI and algorithms to crack the code of human creativity in art. Researchers at a Rutgers University art and AI-focused lab are attempting to create a similar system that will incorporate more novelty into its output rather than just an amalgam of the images it’s shown. Obvious also said it was inspired by Barrat, who has used similar generative networks to paint landscapes, portraits and nudes with an abstract touch as well as AI-generated, 3-D-printed sculptures.

@patrickkulp Patrick Kulp is an emerging tech reporter at Adweek.