From AI use to AI-enabled loafing: a reliance-centred account in hybrid intelligence

Generative AI is increasingly embedded in work-related tasks, raising questions about whether AI-assisted collaboration sustains or weakens meaningful human contribution. This study examines AI-enabled loafing and tests a reliance-centred conditional process model linking AI use frequency, trust in AI, AI reliance, and acceptance of AI labour displacement. Survey data from 404 respondents in China were analyzed using latent-variable structural equation modelling with WLSMV, controlling for employment status and AI literacy. AI use frequency was positively associated with trust in AI, trust was positively associated with AI reliance, and reliance was positively associated with AI-enabled loafing. Although AI use frequency had no significant direct association with AI-enabled loafing, it exhibited significant indirect associations through AI reliance and sequentially through trust and reliance. Acceptance of AI labour displacement strengthened the association between reliance and loafing, as well as both conditional indirect pathways involving reliance. These findings identify AI-enabled loafing as a reliance-centred and normatively conditioned risk. What matters is not simply how often AI is used, but the extent to which users come to rely on it and regard the substitution of human labour as acceptable.