Generative AI in Higher Education: Global Trends, Challenges, and Policy Implications

Abstract

Artificial intelligence has moved from a peripheral tool to a routine feature of student life in higher education. This paper examines how generative AI, and ChatGPT specifically, is used across global and regional student populations, and what challenges accompany that use. Drawing on secondary analysis of four public survey datasets, including a global sample of 23,218 students across 110 countries (Ravšelj et al., 2025) and three available regional datasets from Bangladesh, Colombia, and Ecuador, the study addresses three research questions concerning prevalence and institutional readiness, task level application and regional variation, and the ethical and institutional challenges tied to AI adoption. Results show that 71.4% of the global sample had used ChatGPT, primarily for brainstorming, summarizing, and research assistance, though usage intensity and institutional policy awareness varied meaningfully by region. The three regional datasets converge on a common theme even though each measure it differently: a substantial share of students report reduced verification behavior and self-perceived overreliance on AI tools, a pattern this paper terms experienced dependency to distinguish it from the abstract ethical concern measured by the global dataset. The paper argues that adoption, ethical concern, and experienced dependency function as related but distinct dimensions rather than opposite poles of a single spectrum, and that institutions should track them independently. Implications for policy, faculty development, and future research are discussed.