Are effects of task-related stressors and academic self-management on college students' AI stickiness consistent across the DeepSeek transition?

Steeped in outcome-driven educational ecosystems, college students, especially the first-years facing the transition from high school to university with a surge in unprecedented academic demands, may encounter diverse academic stressors and rely on personal regulatory resources to cope. This challenging transition offers a unique window to investigate how students cope with perceived academic demands and their reliance on self-management skills in the evolving landscape of generative Artificial Intelligence (AI), particularly the changes due to the release of DeepSeek R1. Guided by Conservation of Resources (COR) theory, this study examines the consistency of indirect effects of task-related academic stressors and academic self-management on students' AI engagement via a resource caravan reflected by user-stickiness factors across the DeepSeek transition. Using a repeated cross-sectional design, 640 first-year undergraduates completed surveys across two waves with multi-group SEM analyses testing whether indirect effects differed significantly (Hypothesis 1) or remained consistent (Hypothesis 2). Among the studied indirect effects, only academic self-management consistently fostered stickiness via user satisfaction across two waves, supporting Hypothesis 2, while the effects of academic stressors varied by context at a liberal alpha level of 0.10, providing only partial and preliminary support for Hypothesis 1. Since samples from two waves differed substantially in gender composition and other time-varying contextual factors, the observed differences cannot be attributed solely to the DeepSeek release; they may reflect broader sample and environmental changes. Conversely, the self-management pathway did not differ significantly between two waves involving these compositional and contextual differences. The findings reveal a consistent yet evolving pattern of dependency on AI as a coping strategy for academic stress, with satisfaction emerging as the key psychological resource driving sustained AI use, especially when students possess strong self-management skills.