How We've Taught Algorithms Identity

Race and gender have long sociopolitical histories of classification in technical infrastructures-from the passport to social media. Facial analysis technologies are particularly pertinent to understanding how identity is operationalized in new technical systems. In this talk, Morgan Klaus Scheuerman will present the results of two works on how identity is conceptualized in computer vision. First, "How Computers See Gender," an analysis of how gender is concretely conceptualized and encoded into commercial facial analysis and image labeling technologies available today. Second, "How We've Taught Algorithms to See Identity," how race and gender are defined and annotated in image databases used for facial analysis. The synthesis of these works points towards a need for addressing the discursive construction of identity categories in computer vision, and technologies more broadly.