Automated Decision-Making Systems: How the Accounting Profession Can Contribute Toward Ethical and Responsible Practices
通过对美国、新加坡和印度尼西亚13位高级会计专业人士的访谈，发现行业在自动化决策风险管理中侧重合规和通用治理，但缺乏本地化技术工具和利益相关者反馈，并基于罗尔斯正义原则提出改进框架。
SYNOPSIS The rapid adoption of AI-enabled automated decision-making (ADM) in accounting raises ethical risks around privacy, bias, and accountability. Although prior literature and regulatory frameworks have addressed these issues conceptually, little is known about real-world mitigation strategies deployed by practitioners. Drawing on 13 semistructured interviews with senior accounting professionals from banking, public accounting, finance, IT, retail, and health care in the United States, Singapore, and Indonesia, this study reveals cross-country and cross-industry patterns and gaps in ADM risk mitigation. Our findings show that practitioners emphasize regulatory compliance and general governance (e.g., data classification and training), but they show limited use of localized technical tools (e.g., datasheets, XAI, model and system cards) and elicit minimal stakeholder feedback (especially concerning impacts on less-advantaged groups). Applying Rawls’ distributive justice principles highlights these gaps and also offers a novel assessment lens. We propose a practical Rawlsian-inspired framework to guide accountants toward more equitable ADM practices. Data Availability: Data are available from the authors upon reasonable request, subject to confidentiality constraints. JEL Classifications: M15; M41; M48.