Machine Learning and Postcolonial Critique: Homologous Challenges to Sociological Notions of Human Agency

Abstract

This article discusses two seemingly unrelated but homologous challenges to established sociological thinking, namely machine learning technologies and postcolonial critique. Both of these confront conventional human-centric sociological notions. Where the rise of machine learning should prompt sociologists to take the agency of nonhuman systems seriously, postcolonial critique challenges the idea of Eurocentric human agency. I discuss whether this dual agency challenge can be addressed through Latour’s actor-network theory and Luhmann’s sociological systems theory – both of which explicitly aim to transcend classical human-centric approaches. I argue that Latour’s work can align with postcolonial sociology. However, despite broadening the notion of agency, his actor-network concept remains strongly human-centric. It merely expands the range of actors with which humans engage rather than analysing interactions among nonhuman actants, such as machine learning systems. In contrast, such interactions can be understood through Luhmann’s theorisation, which, however, can be subjected to postcolonial critique.