The Color of Statistics: Race as Statistical (In)visuality

Statistics is interrogated in relation to the program of eugenics via close analysis of its figures of principle component and linear discriminant analysis. These techniques are shown to operationalize race in and for machine learning. They contour and distribute data so that whiteness becomes the normative “color line,” as Saidiya Hartman terms it, of computational experience. This chapter asks: what would it take to artfully prize open AI’s racist aesthetics and allow other colors to shade computational experience? Stephanie Dinkins’s art offers a differently colored AI. She refashions AI by conjoining its operations with Black life, and by engaging Black and people of color’s AI communities. While this chapter delves into machine learning’s deep racism, it likewise shows how Dinkins’s work with AI offers a novel rendering where contingency and difference become possible.