The Zizi Show is an interactive drag cabaret that uses deepfake, an artificial intelligence technique that can generate synthetic images. The project explores the intersection between performance drag, queer identities and the ways in which artistic practices can demystify artificial intelligence.
The show is presented as an interactive show in which a series of deepfake drags perform various songs. Jake Elwes and his team worked with a group of 13 drag queens and kings of various gender and sexual orientations from the London scene to create each character. By recording their performances, they compiled a video database which they used to train an artificial intelligence system capable of creating synthetic images. However, instead of chasing the photorealistic, seamless finish that tech companies and computer scientists often aspire to, the project explores the fractures that come from trying to represent impossible poses or multiple identities, as well as the production processes and aesthetics that emerge from them.
The collaborative creation of deepfakes as a celebration and recognition of identity sets up resistance to oppressive, exploitative practices prevailing through porn or fake news. Just as the drag community questions prevailing notions about gender, The Zizi Show questions and proposes an alternative to the tendency that artificial intelligence systems have towards normativity.