It is part of a project by a Paris-based collective called Obvious, whose aim is to make people think: "Is an algorithm capable of creativity?"
The paintings were created with a two-part algorithm called Generative Adversarial Networks (GANs). First, the Obvious team fed approximately 15,000 portraits, all finished between 1300 CE and 1900 CE, into the algorithm. Next, the first part (the Generator) went about creating its own masterpieces, Edmond De Belamy included. Then, it was the job of the second section of the algorithm (nicknamed “the Discriminator”) to determine whether or not the piece of art was human-made or machine-made.
"The aim is to fool the Discriminator into thinking that the new images are real-life portraits. Then we have a result," Hugo Caselles-Dupré from Obvious told Christie’s.
As you might be able to tell from the above painting, it’s not perfect.
“It is an attribute of the model that there is distortion,” Caselles-Dupré continued. “The Discriminator is looking for the features of the image – a face, shoulders – and for now it is more easily fooled than a human eye.”
All the money raised at Christie's, they say, will go back into the algorithm to improve its artistic prowess. Hopefully, the next round of paintings look a little less alien.
It's not the first time AI has been getting busy in the art room. Last year, an algorithm was programmed to make abstract art – and it did it so well, even the critics were fooled. AI poetry, on the other hand, has some way to go.