The Use of Generative AI in Higher Education Student Assessments: A Synthesis of Empirical Studies

Abstract

Generative artificial intelligence (GenAI) has become increasingly common among higher education students as an aid for completing assessments. Although previous studies have examined the implications of GenAI use in assessments, there is a need to synthesise existing findings to better understand how students use GenAI and how it affects their experiences. This study aimed to synthesise empirical research on higher education students’ use of GenAI in assessments. Guided by the Population, Exposure, and Outcome (PEO) framework, the study adopted a hybrid deductive and inductive codebook thematic analysis of 14 empirical studies identified through a search of the Education Resources Information Center (ERIC) database. The analysis identified four main themes: uses of GenAI in assessments, perceived benefits of GenAI, challenges and limitations of GenAI, and academic integrity and ethical implications of GenAI. The findings suggest that students use GenAI for writing and content development, research and information support, idea generation and planning, and feedback and assessment support. Participants also perceived several benefits associated with GenAI use, including improved assessment performance, increased efficiency, and enhanced learning support. However, the use of GenAI also presents challenges and limitations, including ineffective use, concerns about accuracy and reliability, and effects of over-reliance on learning skills. In addition, participants expressed concerns about plagiarism and academic misconduct, fairness and academic equity, and the need for transparency and ethical GenAI use. The findings suggest that effective and responsible GenAI use requires institutional guidelines and adequate proficiency in GenAI competencies. This study contributes to the growing literature on GenAI in higher education assessments and provides insights for students, educators, and policymakers on the integration of GenAI into assessment designs.