Mar 6
News
An artwork created by AI sold for £40,000 at Sotheby’s, failing to generate the fervor that propelled another AI work to sell for 40 times its estimate last year.
Mario Klingemann, Memories of Passersby I, 2018. Sold for £40,000. Courtesy Sotheby's.

Mario Klingemann, Memories of Passersby I, 2018. Sold for £40,000. Courtesy Sotheby's.

Sotheby’s sold its first-ever work created by artificial intelligence on Wednesday during its day sale of contemporary art in London. Titled Memories of Passersby I (2018), the work by Mario Klingemann generates an infinite, shifting array of portraits on two large digital displays. It hammered to an online bidder at £32,000 ($42,000), according to artnet News, a result that put it on the low end of its pre-sale estimate of £30,000 to £40,000; with fees, the price was £40,000 ($52,634).

The result was a far cry from Christie’s first test of the auction market for AI art last year, when it a sold work by the French collective Obvious for $432,500, or more than 40 times its high estimate of $10,000. The portrait of a non-existent nobleman was generated through a comparison with 15,000 portrait paintings spanning the 14th to the 20th centuries.

Mario Klingemann, Memories of Passersby I, 2018. Sold for £40,000. Courtesy Sotheby's.

Mario Klingemann, Memories of Passersby I, 2018. Sold for £40,000. Courtesy Sotheby's.

The Obvious work sold by Christie’s was promoted far and wide as the first artificial intelligence artwork ever to come to auction, resulting in ample media coverage—and an attribution controversy over the code used to generate the work. By comparison, pre-sale anticipation for Klingemann’s piece at Sotheby’s was more subdued, despite the work’s far more complex technological makeup, perhaps explaining in part the more modest sale result.

Klingemann’s work, which is from an edition of three with two artist’s proofs, uses multiple Generative Adversarial Networks (GANs) to generate portraits in real time on-screen, with the image shifting before the viewer. The GANs generate images based on photographs of portrait paintings spanning the 17th century to the 19th century.

Klingemann called the neural networks involved in the work “the brushes that I’ve learned to use,” in a press release announcing the consignment in February.