Zizi is a procession of faces of drag artists in constant transition generated from data using machine learning(1). Drag is a queer performance form that celebrates and challenges gender and otherness. It's loud, bold and beautiful. Zizi’s gender, identity, race, whether they are  real or artificial, is all uncertain.

Zizi tackles head-on the lack of representation in the training datasets  often used by facial recognition systems. Zizi disrupts these systems by  retraining them with the addition of drag and gender fluidity. This causes  the weights in the neural network to shift away from the normative identities  it was originally trained on.

The work is a celebration of difference and ambiguity, which invites us to  reflect on bias in our data driven society. If AI holds a mirror up to  society, then Zizi applies the makeup.

1) A Style-Based Generator Architecture for Generative Adversarial Networks  (2019)