Auditing Bias in Facial Recognition

This chapter examines the complexities of auditing facial recognition technologies, focusing on performance discrepancies across demographic groups. It highlights the Gender Shades project and its findings on the biases faced by darker-skinned females, stressing the need for robust evaluation methods that reflect intersectional identities. The discussion advocates for a regulatory framework and proposes a temporary moratorium on these technologies to address ethical concerns and spur necessary reforms.