Enhancing visible learning in higher education through transparent and responsible AI: an empirical model based on cognitive load and creative self-beliefs

The increasing integration of Generative Artificial Intelligence (GAI) in higher education has created new opportunities for enhancing learning visibility and instructional transparency. Yet, AI-assisted learning processes often remain opaque, leaving educators uncertain about how AI shapes students’ reasoning, creativity, and ethical awareness. This study seeks to advance visible learning by designing a transparency-oriented AI-supported environment that systematically captures student–AI interaction trajectories within a project-based course. Drawing on Cognitive Load Theory (CLT), Creative Self-Beliefs (CSB), and Responsible AI (RAI) frameworks, the study examines how structured GAI support and perceived transparency relate to learners’ perceived cognitive load and creative self-beliefs. Using behavioral learning analytics and Partial Least Squares Structural Equation Modeling (PLS-SEM), results indicate that GAI support is positively associated with perceived cognitive load reduction both directly and indirectly through strengthened CSB. Furthermore, AI transparency moderates these relationships, such that structured and explainable AI interactions amplify learners’ reflective engagement. The findings extend CLT to complex decision-making contexts, position CSB as a psychological mechanism in AI-supported learning, and conceptualize Responsible AI as a pedagogical design condition rather than merely an ethical add-on. Practical implications are provided for educators seeking to balance cognitive support, creative empowerment, and accountability in AI-integrated higher education environments.