The picture generated by the neural network went under the hammer for $ 432,500

“Edmond de Belami, from the family of de Belami. A controversial neural network, canvas printing, 2018. Signed by the loss function of the GAN model by the ink publisher, from a series of eleven unique images published by Obvious Art, Paris, in an original gilded wooden frame. ” - this is the description of lot 363 , which was sold yesterday at Sotheby’s for $ 432,500.

The total cost of the lot was unexpected for the auction house experts, who initially estimated the picture in the range of $ 7,000 to $ 10,000. But this is also a good salary for a neural network programmer, as it is capable of generating thousands of such pictures.

This work was put up for auction by a group of French entrepreneurs Obvious. The lot went to an anonymous buyer, who participated in the auction by telephone from France.

Art is a good and potentially very profitable application of neural networks: “We would like to thank the AI ​​community, especially those who pioneered this new technology, including Iain Goodfellow, the creator of the GAN algorithm, who inspired the name of the series Famille de Belamy, and the artist Robbie Barrata, who had a great influence on us - saidrepresentatives of the Obvious group. “This is an exciting moment, and we hope that this sale will draw attention to the amazing work our predecessors and colleagues have done.”

Robbie Barrath himself on Twitter provides some examples of great art that is generated by well-trained adversary neural networks.


The programmer and artist Robbie Barrath is a pioneer in the generation of painting using neural networks. By the way, after the auction, he stated that the painting “Edmond de Bellamy” was generated using his program code, which he published in the public domain more than a year ago.


The Obvious group admitted using Barrat’s code, but changed it.

Other modern software artists, including Mario Klingemann and Anna Ridler, use neural networks in painting .