Task demands and performance beliefs: disentangling task importance and complexity in students’ continuance intention toward generative AI

Abstract

Understanding sustained engagement with generative AI (GAI) tools is increasingly important in higher education. Prior continuance research has emphasized performance-oriented beliefs but has less often differentiated task-related conditions. Drawing on complementary propositions from Task–Technology Fit (TTF) and the Technology Acceptance Model (TAM), this study distinguishes task importance from perceived task complexity and examines their direct and indirect associations with students’ continuance intention through perceived performance. Survey data from 199 undergraduate students were analyzed using Hayes’ PROCESS macro with 5,000 bootstrap resamples; adopter-only analyses excluded 12 respondents reporting no prior AI learning experience. Technology characteristics were indirectly associated with continuance intention through perceived performance. Task importance also showed a significant indirect association through perceived performance, while its direct association in the full sample was attenuated in the adopter-only subsample. Task complexity showed no significant indirect association, and its negative direct association in the full sample was likewise attenuated among adopters. Thus, the main indirect associations were retained after excluding non-users, whereas the direct-effect patterns were less stable. Because the study is cross-sectional and task importance and task complexity were measured with psychometrically modest two-item scales, the findings should be interpreted as associative and preliminary. They suggest that distinguishing task importance from task complexity may improve performance-centered accounts of GAI continuance, while workload-related explanations require direct measurement in future research.