From Relational Personhood to Genealogies of Illusion

Abstract

This article reframes the debate on artificial intelligence personhood by arguing that relational accounts are insufficient for a critical understanding of how AI comes to appear as a social subject. It claims that relational theories, while rejecting intrinsic properties and emphasizing interaction, remain limited because they treat social ascription as an empirical given rather than interrogating the conditions that make such ascriptions possible. The article instead proposes a genealogical approach that identifies illusion (understood as a structural feature of mediation rather than a moral failing) as central to the intelligibility of AI personhood. It shows how such illusion emerges through interactional framing, historical techniques of deception, and political-economic processes of labor abstraction. The article further argues that media representations constitute a key site where this illusion is produced and stabilized. Through visual, textual, and sonic examples, it demonstrates how human-like cues and machine-like cues operate together to generate a socio-technical artificial subjectivity that is simultaneously trustworthy and authoritative. It concludes that AI personhood raises not only moral or legal questions about artificial agents, but also political questions about who is empowered, excluded, or rendered passive through the very illusion that AI can function as a person.