Motifsnap

Stream of AI art

It’s quite simple to teach a computer to produce works of art. You provide them with data, and then they try to find patterns within that data and recreate those results. Artificial intelligence systems, meanwhile, are untiring in their output of pictures. German AI artist Mario Klingemann believes the hardest aspect is figuring out what to do with all of the information.

Image 38
Mario Klingemann’s Memories of Passersby I, which will be auctioned at Sotheby’s on March 6th.
 Image: ONKAOS

This possibility is what makes the task both exciting and challenging to me. Choosing just one item [that the algorithm generates] seems incorrect. Although it may generate several pictures, the real enchantment is in seeing it at work.

AI art at auction

This is the same same method that was used to sell the second piece of AI art at a prestigious auction. Both of Memories’ displays use artificial intelligence to create a new portrait every few seconds. All the pictures are different and they change into the next one without any noticeable transitions. The effect is similar to that of seeing a lava lamp composed entirely of human features.

After years of development, these sales show that AI art is finally making its way into the mainstream art market. However, in doing so, it raises interesting considerations regarding artistic expression and originality. When it comes to art and technology, what kind of connection do they have? Can a computer software ever be considered creative?

The two systems exchange images back and forth until the discriminator can no longer detect the difference between the bogus data and the real training material. In the case of Memories, there was a large collection of portraits from the 17th, 18th, and 19th centuries, all hand-picked by Klingemann. He also modifies the network’s strictness and the particular features it approves of, influencing the results it generates.

Image 40
Sample faces generated by Klingemann’s GAN.
 Image by Mario Klingemann

In art school, I learned that it takes a lot of calm observation time before your brain can process what you’re seeing. The visuals put the viewer inside the AI’s head, which helps them understand what it wants to do. When compared to spending hours upon hours upon hours upon days upon days upon days standing still and painting, it is easy to see why a simple printout would not do justice to the full scope of the work.

Shopping cart close