Race and AI: the Diversity Dilemma

Stephen Cave

This commentary is a response to 'More than Skin Deep' by Shelley M. Park (Park, More than skin deep: A response to "The Whiteness of AI", Philosophy & Technology, 2021), and a development of our own 2020 paper 'The Whiteness of AI'. We aim to explain how representations of AI can be varied in one sense, whilst not being diverse. We argue that Whiteness's claim to universal humanity permits a broad range of roles to White humans and White-presenting machines, whilst assigning a much narrower range of stereotypical roles to people of colour. Because the attributes of AI in the popular imagination, such as intelligence, power and passing as human, are associated by the White racial frame with Whiteness, AI is cast predominantly as White. Following Sparrow (Science, Technology, & Human Values 45:538-560, 2020), we suggest this presents a dilemma for those creating or representing AI. We discuss three possible solutions: avoiding anthropomorphisation, explicitly critiquing racial role-typing, and representing powerful AI as non-White.