THE IMPACT OF GENERATIVE AI DEPENDENCY ON LEARNING OUTCOMES: THE MEDIATING ROLE OF COGNITIVE AGILITY AMONG UNIVERSITY STUDENTS

Abstract

The educational paradigm in higher education has been completely changed by generative artificial intelligence (GAI) technologies like ChatGPT, Gemini, and Copilot. These tools offer students unparalleled access to knowledge and learning help while also enhancing the teaching process with creative methods. Although helpful, there are worries about how students' reliance on AI-generated work could affect their learning outcomes. The present study examined the mediation effect of CA (Cognitive Agility) on the relationship between GAD (Generative AI Dependency) and LO (Learning Outcomes) among the students of being at the university level. The research suggested that overreliance on generative AI could reduce students' cognitive load relative to their own thinking, but cognitive agility may enable them to adapt, critically process, and effectively use information created by the generative AI. The research method adopted was quantitative cross sectional and the data obtained from 400 respondents who are students at the universities in the city, using stratified random sampling technique. The data were collected with three standardized self-report instruments, namely the Generative AI Dependency Scale (8 items), the Cognitive Agility Scale (10 items) and the Learning Outcomes Scale (9 items), rated on a 5-point Likert scale. Cronbach's α for the scales were satisfactory with ranges between .87 and .91. The relationship of the study variables was examined using Structural Equation Modeling (SEM) in the form of direct and indirect relationships. The results showed that Generative AI Dependency had a significant impact on Learning Outcomes and Cognitive Agility, and that Cognitive Agility had a significant impact on the Learning Outcomes. Additionally, a partial mediation analysis was performed in between Generative AI Dependency and Learning Outcomes with Cognitive Agility. By identifying the psychological domains that impact student learning with the use of AI, this study contributes to the existing research on AI in education. The findings are relevant for learning designers, policy makers and higher educational institutions who aim to develop meaningful and effective learning in today's technological environment