ARTificial Creativity and Invention

Artificial intelligence techniques have experienced a renaissance in recent years with neural networks, a type of algorithm that is (loosely) modeled after connectionist models of the brain. In this exhibit, we explore the use of generative adversarial networks (GANs) which learn to create visual media by training on large datasets of human-generated images, and then output novel images when fed a random vector input, generating the image ex nihilo. Experience the progression of images generated by GANs from their early nascent stages as pixelated creatures through their turbulent adolescent times before reaching a tentative mature adulthood that you see at the end.