Asymmetric Dependency and the Algorithmic Figuration

Towards a Figurational Sociology of Artificial Intelligence

Abstract

This article argues that AI-governed systems, and large language models in particular, introduce a novel dynamic into contemporary figurations: one of asymmetric dependency, in which people are materially, and increasingly affectively, dependent upon systems that are not, in turn, affectively dependent upon them. This condition attenuates the affectively reciprocal interdependency on which psychogenetic processes depend in Norbert Elias’s account of the civilising process and gives rise to what we distinguish as behavioural civilising: the shaping of observable conduct through monitored compliance rather than through the internalisation of restraint. Because such systems simulate human exchange whilst stripping it of its negative social consequences, they simultaneously suspend the shame functions through which figurational pressure is ordinarily converted into self-restraint and so exert decivilising pressures within the same figuration. Extended engagement with conversational AI also introduces generative reflection: a mode of exchange in which thought is elaborated and transformed rather than merely returned. We thereby extend the Eliasian framework whilst specifying the conditions under which psychogenetic processes are disrupted and transformed.