不确定性审计任务中人工智能的使用与投资者感知的审计师责任——一项实验研究  

Use of AI in Uncertain Audit Tasks and Investor Perceptions of Auditor Liability——An experimental evidence

作　　者：柳扬 苗靖钰 刘文欢 何亚南 Liu Yang;Miao Jingyu;Liu Wenhuan;He Yanan

机构地区：[1]南开大学商学院,300071 [2]青岛大学商学院,266001 [3]上海立信会计金融学院,201620

出　　处：《审计研究》2025年第2期110-119,共10页Auditing Research

基　　金：国家自然科学基金项目(项目批准号:72272080、72302123和72202204)的资助。

摘　　要：会计师事务所在审计中运用人工智能等先进技术受到广泛关注。基于行为理论和实验研究方法,本文检验了使用人工智能执行错报相关的计量不确定性审计任务,对投资者感知的审计师责任的影响。研究发现,在计量不确定性较低任务下,当审计后仍然被发现存在重大错报时,无论是人类审计师还是审计师使用人工智能执行错报相关审计任务,投资者感知的审计师责任没有明显差异;在计量不确定性较高任务下,相比于人类审计师,当使用人工智能执行错报相关审计任务时,投资者感知的审计师责任更高。此外,还发现在计量不确定性较高任务下,审计方式影响投资者感知的审计师责任的具体路径。研究对监管机构、审计师等都具有重要意义。In recent years, accounting firms have increasingly used advanced technologies such as artificial intelligence in the audit process, drawing much attention to its influence. Based on behavioral theory and experimental research methods, this paper examines the effect of using artificial intelligence to perform a measurement-uncertainty audit task related to misstatements on investors' perception of auditor liability. The study finds that in the task of low measurement uncertainty, when material misstatements are still found after the audit,there is no significant difference in investors' perception of auditor liability, whether human auditors or artificial intelligence is used to perform audit tasks related to misstatements. In the task of high measurement uncertainty,investors perceive auditor liability to be higher in the case of using artificial intelligence to perform audit tasks related to misstatements than that of human auditors. In addition, it also finds the mediator mechanism between audit approach and investors perception of auditor liability when the measurement uncertainty of audit task is high. The study provides important insights for regulators and auditors.

关 键 词：人工智能 计量不确定性 投资者感知 审计师责任 

分 类 号：F239.4[经济管理—会计学]