Deconstructing Whiteness: Visualizing Racial Bias in a Face Recognition Algorithm

Abstract

Deconstructing Whiteness is an art performance which examines the visibility of race through the lens of AI face recognition technology. The artist interacts with an off-the-shelf face recognition algorithm, while specifically examining the confidence level by which it recognizes her as ‘White’. During the performance she changes her appearance by modifying her facial expressions and hair style and by doing so is able to ‘trick’ the algorithm and affect its confidence regarding her racial visibility.