California artist Alexander Reben’s series “amalGAN” uses a type of machine learning, generative adversarial networks, that pits one learning network against another.
Reben uses algorithmic models from Artbreeder, an AI resource and social media site. The AI combines words such as “flower” and “dog,” produces images depicting the resulting nonsense words, and “breeds” two images. Reben looks at the spawn, and another AI program reads his brainwaves and body signals, determining which ones grab him. He sends those — familiar but monstrous, sometimes unnervingly creepy — to a painting workshop in China, where anonymous painters put them on canvas.
January 27, 2022
