Autism and the making of emotion AI: Disability as resource for surveillance capitalism

Abstract

This article uncovers how computers learned to recognize emotion from a neurological difference often conceived of as marked by a difficulty in recognizing it at all: autism. Developers of facial emotion recognition technologies have repeatedly mobilized autism to produce new kinds of knowledge at the interface of machine learning and human feeling, deploying it as a charismatic use-case, as a source for conceptual schemata for how emotion should be made computable, and as a testbed for development. The entanglement of emotion recognition and autism research shows how disability has been transformed into a rhetorical, conceptual, and material resource for the expansion of surveillance capitalism. This history has also underwritten a larger reconceptualization of emotional data, from a neglected variable in human–computer interaction to an exploitable corporate asset, a transformation that has allowed platform users’ emotional lives to be mined for new forms of knowledge, value, and power.