When Convenience Undermines Cognition: Quasi-Experimental Based Evidence of AI Use in the University Students

Associate Professor, Department of Economics, National University of Modern Languages (NUML), Islamabad Reference: This study attempts to assess the impact of students' exposure to Artificial Intelligence (AI) tools, in terms of use of AI tools, on their cognitive learning outcomes. The study uses an original micro dataset of over 2000 University students. The study employs a quasi-experimental strategy to accomplish its objectives by comparing a treatment group (with AI exposure) with a control group (without AI exposure). To establish causality, a difference-in-difference framework has been used. The results indicate a striking and meaningful divergence between the two groups, i.e. the treatment group has lower cognitive learning gains as compared to that of the control group. The results persist when the observable attributes and baseline performance is taken into account. The results suggest that reliance on AI tools may lead to poor cognitive skills development, as far as the studied group is concerned. This study attempts to identify the down-side of AI use, i.e. effortless retrieval and too much reliance on AI generated solutions, and identifies policy implications, particularly related to curriculum design, assessment practices, and AI use governance in higher education. Key Word: AI Use, Cognitive Learning, Quasi-Experiment, University Students, Pakistan