AI-Mediated Critical Thinking: A Double-Edged Sword in Indonesian Higher Education

The proliferation of generative artificial intelligence (AI) in higher education presents a paradoxical challenge: while offering unprecedented access to information, it may simultaneously erode the very cognitive skills it aims to enhance. This study investigates the moderating role of digital literacy on the relationship between AI usage patterns and critical thinking skills among 245 Indonesian university students engaged in online learning. Employing a cross-sectional survey design, data were collected using the Watson-Glaser Critical Thinking Appraisal (WGCTA) and the European Commission's Digital Competence Framework (DigComp 2.2). Multiple regression and Partial Least Squares Structural Equation Modeling (PLS-SEM) analyses revealed that AI usage intensity negatively predicted critical thinking (β = -0.24, p < 0.01). However, this relationship was moderated by digital literacy (β = 0.22, p < 0.01), with high digital literacy students demonstrating positive associations between AI use and critical thinking (β = 0.15, p < 0.05). In contrast, low digital literacy students showed stronger negative effects (β = -0.38, p < 0.001). Cluster analysis identified three distinct user profiles: selective users (35.1%) with the highest critical thinking scores (M = 35.8), instrumental users (41.6%, M = 32.1), and dependent users (23.3%, M = 27.4). These findings suggest that AI's impact on critical thinking depends on users' digital competencies, underscoring the need for integrated digital literacy curricula to harness AI's potential while mitigating cognitive offloading risks.