The Impact of ChatGPT’s Advice on Professional Judgment Related with Accounting and the Role of Accounting Education

This study examines the impact of generative artificial intelligence (GenAI) advice on ethical judgment in accounting and auditing, particularly in comparison to conditions involving no advice and/or advice from a human auditor. We employ an experimental design, in which the independent variable consists of five distinct advice conditions, varying by source (GenAI or human auditor) and by framing context (deontological or consequentialist). We simulate an experimental scenario in which financial controllers either consider or disregard advice from external auditors during audit negotiations. In this context, accounting majors are used as proxies for financial controllers working in the finance department. To assess how these conditions influence ethical judgment, we conducted several logistic regression analyses. Participants were accounting majors recruited from seven leading universities in Japan. The results provide compelling evidence that receiving advice, whether from GenAI or a human auditor, increases the likelihood of making a deontological judgment (i.e., a judgment grounded in moral duties, rules, or principles), compared to receiving no advice. This suggests that the impact of GenAI advice closely mirrors that of human auditors. Further analysis revealed that among participants receiving GenAI advice, those exhibiting a higher degree of algorithm appreciation (i.e., the tendency to perceive algorithmic advice as more credible) were significantly more inclined to make deontological judgments. Moreover, when comparing advice from GenAI and human auditors directly, the findings confirm that the effects of moral advice from both sources were remarkably similar. However, contextual factors played a critical role: participants with greater familiarity with consolidated accounting standards (KNOW) and those who had completed internships at accounting firms (INTER) were more likely to make deontological judgments—even when exposed to consequentialist advice (i.e., advice oriented toward maximizing overall welfare). These findings offer important implications for accounting ethics education and highlight the need to consider individual differences and contextual factors in algorithm-assisted decision-making. The study also acknowledges several limitations and suggests directions for future research.