Key points are not available for this paper at this time.
Amid the accelerating diffusion of generative AI (GenAI), this study investigates generative AI anxiety and its impact on deep engagement in the post-adoption stage. Employing a sequential mixed-methods design in China from October to November 2025, we first proposed the construct of deep use intention and utilized interviews (N = 16) via purposive sampling and procedural grounded theory to map a multidimensional landscape of anxiety toward GenAI among young adults, revealing a shift in the formation mechanism from sociotechnical blindness to sociotechnical awareness. Subsequently, we conducted a national survey (N = 899) employing random and snowball sampling. Data analysis via partial least squares structural equation modeling (PLS-SEM) using SmartPLS 4 revealed that: (1) deep use intention was primarily driven by factors other than AI anxiety; (2) while aggregate AI anxiety exerted no significant effect on deep use intention, its internal dimensions exhibited divergent impacts; and (3) AI literacy was validated as a necessary threshold for deep use intention but failed to moderate the impact of AI anxiety, reflecting a potential homogenization of psychological responses. This study advances the understanding of user behavior and the potential AI outcomes divide in the post-adoption stage.
Zhao et al. (Thu,) studied this question.