International Journal of
Information and Education Technology

Abstract

The integration of Generative Artificial Intelligence (GenAI) in Science, Technology, Engineering, and Mathematics (STEM) education has created an “Efficiency Paradox”, where students may rely on algorithms not out of incompetence, but to manage increasing academic workloads. Distinct from previous research that predominantly examines AI through the lens of academic integrity or general acceptance, this study investigates the psychological mechanisms linking student engagement to AI dependence and output productivity. This study empirically modeled the dual pathways of student engagement and AI dependence on output productivity. Employing a descriptive-predictive correlational design, data were collected from 915 undergraduate STEM students at a state university in the southern Philippines. Partial Least Squares Structural Equation Modeling (PLS-SEM) and Importance-Performance Map Analysis (IPMA) were utilized to analyze the relationships between digital literacy, perceived usefulness, self-efficacy, motivation, time management, AI dependence, and productivity. The structural model explained 75.2% of the variance in productivity (R2 = 0.752). The findings revealed that Time Management was the single strongest driver of both AI Dependence (β = 0.409) and Productivity (β = 0.605), suggesting that dependence is often a strategic adaptation for efficiency. Conversely, Intrinsic Motivation acted as a buffer, significantly reducing reliance on AI tools (β = −0.460). IPMA results highlight that while students possess high digital literacy, it has low impact on productivity compared to regulatory skills. The study concludes that higher education policy at the institutional level should shift focus from basic AI literacy to fostering self-regulation and time management to prevent pathological dependence.