When Control is Lost but Meaning is Gained: A Moderated Mediation Model of Digital Agency and AI Adoption in Higher Education
The rapid expansion of artificial intelligence (AI) in higher education has intensified concerns regarding autonomy, control, and ethical engagement with digital technologies. While prior research has primarily addressed AI adoption through acceptance- and performance-oriented frameworks, less is known about how students’ experiences of digital agency shape the psychological mechanisms underlying their intention to use AI. Grounded in digital agency theory and expectancy–value perspectives, the present study examines a moderated mediation model explaining students’ intention to use AI in higher education. Using survey data from 673 university students, a conditional process analysis (PROCESS Model 59) was conducted to test whether sense of negative agency (SONA) predicts intention to use AI indirectly through perceived value of AI, and whether this pathway is conditioned by sense of positive agency (SOPA). Performance control was examined as a parallel mediator to assess the role of self-regulated learning mechanisms. Results revealed a significant moderated mediation effect. Specifically, perceived value of AI mediated the relationship between negative agency and intention to use AI only at moderate to high levels of positive agency. When SOPA was low, negative agency was unrelated to perceived value and intention. In contrast, at higher levels of SOPA, negative agency was positively associated with perceived value of AI, which in turn strongly predicted intention to use AI (b = .83, p < .001). Performance control did not emerge as a significant mediator, indicating that AI adoption decisions were not driven by self-regulated performance mechanisms. These findings suggest that students’ engagement with AI is guided primarily by value-based cognitive evaluation rather than by performance regulation. Experiences of reduced control do not necessarily inhibit AI adoption; instead, when integrated through positive agency, they may foster reflective meaning-making and intentional AI use. The study highlights the ethical relevance of digital agency in AI adoption and underscores the importance of agency-aware approaches to AI integration in higher education.
Paper
Full text
When Control is Lost but Meaning is Gained: A Moderated Mediation Model of Digital Agency and AI Adoption in Higher Education
Semantic Scholar · 2026
Abstract
The rapid expansion of artificial intelligence (AI) in higher education has intensified concerns regarding autonomy, control, and ethical engagement with digital technologies. While prior research has primarily addressed AI adoption through acceptance- and performance-oriented frameworks, less is known about how students’ experiences of digital agency shape the psychological mechanisms underlying their intention to use AI. Grounded in digital agency theory and expectancy–value perspectives, the present study examines a moderated mediation model explaining students’ intention to use AI in higher education. Using survey data from 673 university students, a conditional process analysis (PROCESS Model 59) was conducted to test whether sense of negative agency (SONA) predicts intention to use AI indirectly through perceived value of AI, and whether this pathway is conditioned by sense of positive agency (SOPA). Performance control was examined as a parallel mediator to assess the role of self-regulated learning mechanisms. Results revealed a significant moderated mediation effect. Specifically, perceived value of AI mediated the relationship between negative agency and intention to use AI only at moderate to high levels of positive agency. When SOPA was low, negative agency was unrelated to perceived value and intention. In contrast, at higher levels of SOPA, negative agency was positively associated with perceived value of AI, which in turn strongly predicted intention to use AI (b = .83, p < .001). Performance control did not emerge as a significant mediator, indicating that AI adoption decisions were not driven by self-regulated performance mechanisms. These findings suggest that students’ engagement with AI is guided primarily by value-based cognitive evaluation rather than by performance regulation. Experiences of reduced control do not necessarily inhibit AI adoption; instead, when integrated through positive agency, they may foster reflective meaning-making and intentional AI use. The study highlights the ethical relevance of digital agency in AI adoption and underscores the importance of agency-aware approaches to AI integration in higher education.