The Void of Surveillance: Machine Learning, Psychoanalysis, and the Misreading of Desire

Surveillance studies has long drawn on Michel Foucault’s (1977) panopticon and Gilles Deleuze and Félix Guattari’s (1987) assemblage to analyse how surveillance functions and its many societal effects. More recently, there has been a particular emphasis on how profiles, otherwise known as data doubles, have played an increased role in surveillance using artificial intelligence. While this scholarship has explored how data are gathered and analysed, this article problematises the ability for the data flowing from humans to machines to be so revelatory of subjectivity. In doing so, this article considers how adopting a psychoanalytical notion of desire, drawn from Jacques Lacan (2007), problematises the assumption that data stemming from outward affirmations of subjectivity are an accurate reflection of who we are. If desire is bound to lack, instead of being an active, positive force, this raises a host of questions about how surveillance is confronted by cuts, blockages, absences, and exclusions when trying to make sense of an individual. As this article argues, this gap of knowledge between profiles and individuals is not because there is a flaw within surveillance but because surveillance itself depends on a constitutive void: the necessary gap of knowledge between data and desire, in light of machines only ever being able to interpret humans as machines. The void of surveillance is not what merely sustains surveillance, as needing to know more to fill a gap of knowledge, but sustains the necessary failure of total surveillance’s goal of knowing everything there is to know about an individual, given there is always more to know. This article shows how paying attention to this void offers new insights into surveillance’s misunderstanding of subjectivity.