The Impact of Generative AI Tools on Academic Performance in Higher Education: A Systematic Review (2023–2025)

Abstract

This paper examines how generative AI tools—such as ChatGPT, GitHub Copilot, and related large language models—are influencing academic performance in higher education. We reviewed empirical studies published between 2023 and 2025 and followed PRISMA guidelines to search major open-access databases, including ERIC, DOAJ, arXiv, MDPI, and Semantic Scholar. Our review focused on research that reported measurable outcomes such as GPA, exam scores, writing quality, and coding accuracy. In total, 14 studies met the inclusion criteria, representing a range of countries, disciplines, and instructional contexts. The results are far from uniform. Several studies reported gains in writing, coding, and task efficiency when generative AI was used in a structured and purposeful way. Others, however, found little improvement or even lower performance when students relied heavily on AI without guidance. Taken together, the evidence shows that AI’s impact on learning is highly varied and strongly dependent on context. This underscores the need for deliberate integration and for more sustained, data-driven research as these tools continue to evolve.