ARTIFICIAL INTELLIGENCE TOOLS AND STUDENT LEARNING IN HIGHER EDUCATION: A STRUCTURED REVIEW OF BENEFITS, RISKS, AND IMPLEMENTATION CONDITIONS

Abstract

Background. Generative artificial intelligence (AI) tools have entered university classrooms faster than institutions have been able to evaluate them. Objective. This article synthesizes research on how AI tools affect student learning in higher education and identifies the conditions under which they help or harm. Methods. A structured narrative review was conducted following the logic of PRISMA-style screening, with defined search sources, inclusion criteria, and a thematic coding procedure. Results. Four themes emerged: personalized support and feedback, development of higher-order skills, risks to academic integrity and cognitive engagement, and the decisive role of institutional and instructor mediation. Conclusion. AI tools tend to support learning when they are embedded in clear pedagogical design, paired with transparent assessment policies, and accompanied by AI-literacy training for both teachers and students. Without such conditions, they risk encouraging superficial learning and unequal outcomes