The Zizi Show is a online deepfake drag cabaret. During COVID-19 lock down a community of drag queen artists in London were filmed to train a dataset on of bodies. The dataset was used to train a neural network to generate deepfakes. In the interactive part of the project users can choose between drag artists and performance numbers creating a own drag cabaret. The playful approach to deepfakes presented in The Zizi Show is intended to "dissects one of the dominant myths about AI, the notion that ‘an AI’ is a thing we might mistake for a person."(1)
The project recognizes that facial recognition technology "currently have difficulty recognising trans, queer and other marginalised identities, because they are often made by cis white people"(2) and a part of the project is to queer the normative datasets that are used to train various facial recognition models.
Rather than seeing deepfakes as a threat it tries to de mystify AI asking critical questions "whether making deepfakes using queer identities becomes a means of assimilation or inclusivity… or more a techno-activist method of dirtying and obfuscating the systems used to collect data on us." (2) In this context the use of deepfakes becomes something an empowering tool celebrating queerness.
1) https://computervisionart.com/pieces2021/the-zizi-show/
2) https://zizi.ai/about