Addicted to Artificial Intelligence? Unpacking the Role of Attitudes and Agency in AI Addiction through the I-PACE Model Abstract

Artificial intelligence (AI) provides significant benefits but also risks psychological overdependence, termed AI addiction (AIA). Understanding the mechanisms underlying this emerging form of addiction is thus essential. Anchored in the Interaction of Person–Affect–Cognition–Execution (I-PACE) model, we conceptualize general attitudes toward AI (GAAI) as “person-level” dispositions and the sense of agency (SoA) as the “cognitive-affective” response that mediates AIA. Study 1 (NSample 1 = 500; NSample 2 = 896) adapted and validated a concise Chinese General Attitudes toward AI Scale, confirming a stable two-factor structure with good reliability. Study 2 (N = 780) applied this scale to a parallel mediation model linking GAAI to AIA via positive (SoPA) and negative (SoNA) facets of SoA. Results revealed that positive attitudes towards AI (GAAIP) indirectly mitigated addiction by enhancing SoPA and reducing SoNA, whereas negative attitudes towards AI (GAAIN) positively associated with GAA through diminishing SoPA and increasing SoNA. These findings extend the application of the I-PACE framework in the AI field, by demonstrating the association between GAAI and AIA, and the mediating role of SoA. Our findings also have the potential to guide the development of cognitive or attitudinal interventions for AIA.