Psychological absorption and the maintenance of agency in AI-mediated cognition: a relapse prevention perspective—V4

Abstract The rapid integration of generative artificial intelligence into professional, educational, and therapeutic contexts has intensified debates about efficiency, cognitive offloading, and human agency. While existing accounts emphasize functional and performance-related outcomes, they offer limited frameworks for examining how sustained AI use may shape subjective experience over time. This paper proposes a conceptual, hypothesis-generating model of psychological absorption to describe a gradual process in which cognitive, regulatory, and meaning-making functions are increasingly delegated to AI systems. Drawing on relapse prevention theory, the model conceptualizes absorption not as addiction or loss of control, but as a maintenance vulnerability in agency that emerges through repeated, adaptive reliance on external cognitive support under conditions of overload and uncertainty. The paper integrates relapse prevention concepts with selected psychodynamic perspectives to illuminate how containment, authorship, and experiential vitality may be subtly reshaped in AI-mediated cognition. Rather than advocating restriction or abstinence, the framework emphasizes early detection, awareness of micro-delegations, and the restoration of psychological presence. By reframing AI use as a maintenance challenge rather than a pathological dependency, the paper contributes to interdisciplinary discussions on human-AI interaction, subjectivity, and the future of agency in AI-saturated environments.