Student perceptions of generative AI in accounting education: A thematic analysis

Abstract

Generative AI has outpaced pedagogical and ethical guidance in accounting classrooms, leaving students and educators uncertain whether it supports or undermines the discipline's dual emphasis on procedural accuracy and professional judgment. To address this problem, this study examines how 92 business undergraduates perceive generative AI tools, notably ChatGPT, in accounting education, drawing on in-class, handwritten survey responses. Findings show that students view AI as a useful yet uncertain resource: valuing its speed, accessibility, and support while raising concerns over cheating, fairness, unclear institutional policies, and overreliance. Four interconnected themes emerged: Practical Affordances and Engagement, Ethical Tensions and Integrity, Curricular and Institutional Context, and Reflective Learning and Active Engagement. These results suggest student perceptions are shaped less by technology alone than by the pedagogical and ethical conditions of its introduction. Theoretically, the study identifies an augmentation–formation paradox specific to disciplines that combine codifiable procedure with professional judgment, which is advanced tentatively as extending the boundary conditions of paradox theory in education, a proposition that requires further empirical testing. Thematic analysis using a hybrid inductive–deductive coding strategy served as the primary analytic approach, supported by supplementary lexical analyses. The study contributes to management and accounting education by demonstrating that responsible AI integration requires clear guidance, reflective learning, and explicit alignment among AI use, assessment design, and academic integrity.