Beyond ‘good’ or ‘bad’: Investigating trust and techno-resistance in postgraduate students' voluntary use of AI technologies

The rise of generative artificial intelligence (AI) has created both opportunities and tensions in postgraduate education, where voluntary adoption is shaped not only by technical functionality but also by perceptions of trust, content quality, and academic integrity. This study extends technology adoption theories by proposing a dual-path model that distinguishes the drivers of voluntary adoption from those of techno-resistance—two processes often treated as identical in prior research. Using survey data from 170 postgraduate students in Malaysia, the findings demonstrate that functional trust (confidence in AI’s stability and reliability) significantly predicts voluntary usage (β = 0.168, p < 0.001). In contrast, evaluative trust (confidence in AI’s intellectual adequacy and academic validity) does not reduce resistance. This highlights an asymmetry in the role of trust, where technical dependability promotes adoption, but academic credibility does not automatically diminish resistance. The study also introduces the construct of epistemic utility, defined as the perceived richness, relevance, and scholarly value of AI-generated content. Results show that epistemic utility is the strongest predictor of adoption (β = 0.785, p < 0.001), underscoring students’ emphasis on content quality over technical reliability. Moreover, while system reliability reduces techno-resistance (β = −0.176, p = 0.034), adoption and resistance stem from distinct antecedents. Significantly, voluntary usage improves academic performance (β = 0.270) more than resistance hinders it (β = −0.210). Together, these findings advance theory by clarifying trust differentiation and introducing epistemic utility as a critical lens for understanding postgraduate engagement with AI.