In Permutations I explore and exploit the ability of a Generative Adversarial Network (GAN) to become something of a virtual scientist. Photographic representations of objects are melded together to form an image that doesn’t exist, but looks like it could in a near-future environment. Aspects within the image may be familiar, but it’s still inherently alien to what we interpret as a “real” space or object. Each piece is made up of images fed into the network, which looks at these sources and reassembles them into a new image, based on what it’s learned of how things should look. Depending on the image ratios, characteristics of the source objects may or may not be readily apparent. The work is largely a collaboration between human and machine - I control these ratios, editing the “genes” of my pieces, and then allow the GAN to generate what it believes is the resultant image. Together we are determining what of the physical world can be joined, and how those hybrids might appear based on what it’s learned.