This work was created using a generative adversarial network trained on digital images of paintings by Wassily Kandinsky. Two parties are trained simultaneously through an adversarial process. The generator, as the artist, tries to generate authentic images, while the discriminator, as the art critic, learns to distinguish the original from the imitated works.
This was the intial project to experiment with a tool that can be used in future design processes. Design processes like these can be considered computational design, as John Maeda defines it.