Kokoro Tomita

This article theorizes compulsory autonomy as the discursive regime that sustains the cultural fantasy of autonomous Artificial intelligence (AI) by converting structural dependence into fixable anomalies (foreclosure) and rendering constitutive supports unthinkable (erasure). By this, I refer to a discourse governing human institutions and representations, not machines as dominated subjects. Adapting Judith Butler's account of foreclosure and Yu Matsuura's distinction between foreclosure and erasure from asexuality studies, alongside Bernard Stiegler's organological philosophy, I show how dominant AI narratives preempt the legibility of ongoing human and material labor. A brief genealogy traces how autopoiesis traveled from biology and cybernetics into AI rhetoric, where it underwrites the misrecognition of heteronomous systems as self-producing. Illustratively, I read reporting on content moderation and Reinforcement Learning from Human Feedback (RLHF) pipelines alongside Pasquinelli's history of ghost work to demonstrate how "autonomy" is performed atop distributed, often precarious labor. I advance the metaphor of the labored cognitive assemblage, outlining policy, design, and public literacy implications (e.g., disclosure norms and labor standards) that follow when dependence is theorized as intrinsic rather than transitional. Reframing autonomy in these terms redirects debates about AI's social futures toward accountability, maintenance, and care.

This article theorizes compulsory autonomy as the discursive regime that sustains the cultural fantasy of autonomous artificial intelligence (AI) by converting structural dependence into fixable anomalies (foreclosure) and rendering constitutive supports unthinkable (erasure). By this I refer to a discourse governing human institutions and representations, not machines as dominated subjects. Adapting Judith Butler's account of foreclosure and Yu Matsuura's distinction between foreclosure and erasure from asexuality studies, alongside Bernard Stiegler's organological philosophy, I show how dominant AI narratives preempt the legibility of ongoing human and material labor. A brief genealogy traces how autopoiesis traveled from biology and cybernetics into AI rhetoric, where it underwrites the misrecognition of heteronomous systems as self-producing. Illustratively, I read reporting on content moderation and RLHF pipelines alongside Pasquinelli's history of ghost work to demonstrate how "autonomy" is performed atop distributed, often precarious labor. I advance the metaphor of the labored cognitive assemblage, outlining policy, design, and public-literacy implications (e.g., disclosure norms and labor standards) that follow when dependence is theorized as intrinsic rather than transitional. Reframing autonomy in these terms redirects debates about AI's social futures toward accountability, maintenance, and care.