Abstract

Recent socio-technical developments in artificial intelligence (AI) increasingly relocate directive authority into algorithmic infrastructures, positioning predictive systems not merely as auxiliary tools, but as structuring conditions of human action. This paper theorizes this transformation as social role inversion: a structural configuration in which algorithmic systems assume directive, evaluative, and coordinative functions traditionally associated with human authorship, while human actors are repositioned as responsive participants within system-defined parameters. By tracing the historical metaphors of the social organism, extending from Plato’s teleology and Hobbes’s artificial sovereign to La Mettrie’s mechanistic reductionism, the paper reveals a profound normative challenge. The central issue of AI governance is not machine agency, but the relocation and attenuation of accountable authorship under predictive mediation. As authority becomes infrastructural, the attribution of responsibility becomes opaque and legitimacy risks reduction to performance optimization. The analysis culminates in the concept of ontological capture, describing the psychological and structural threshold where algorithmic evaluative logics are internalized within subjectivity itself. The paper provides a unified theoretical framework for institutional reconstruction by outlining conditions, such as revocability and meaningful human control, under which algorithmic coordination can remain compatible with democratic legitimacy.