abstract
- Amid the fast dissemination of Artificial Intelligence (AI) into educational institutions, instead of a polarized viewpoint, either hype-AI or hypo-AI, this innovative work proposes an ethically balanced human-AI pedagogical model. Our research demonstrates how a pedagogically guided generative AI tutor, such as LuzIA, can support self-study processes in higher education while preserving human instructional authority. The study followed a quasi-experimental design involving two groups-experimental (n=30) and control (n=29) -of undergraduate business students. The intervention was implemented in a Mathematics Reasoning course in August-December 2024. Through a mixed-methods approach, we analyzed the differences in average final-test scores as well as students' responses to a Technology Acceptance Model (TAM) questionnaire. The findings suggest design requirements for these tools and highlight their potential to provide flexible and personalized support without overburdening the instructor. Furthermore, students' responses reframe technology acceptance as a function of learning efficacy rather than usability alone.