Theories and Dimensions of Algorithmic Power, between Agency and Domination

Abstract

This article reviews the critical and sociological literature on algorithms and AI in light of anthropocentric social theories of power and its dimensions. Building on the classic conceptual distinction between power to (agency) and power over (domination), I outline two main theorizations of algorithmic agency, which here I call “networked” and “cultural”, and conceptualize four dimensions of algorithmic domination: opaque coercion, computational authority, structural conditioning, and data governmentality.