【学术讲座】生成式人工智能与关键审计事项

发布时间:2026-11-16

Generative AI and Critical Audit Matters

讲座信息 | INFORMATION

讲者:Huimin (Amy) Chen, Associate Professor at the University of Massachusetts

时间:2026年12月3日(周四) 14时30分

地点:同济大厦A座510教室

讲座摘要| ABSTRACT

We investigate how generative artificial intelligence affects critical audit matter (CAM) disclosures and audit procedures. Generative AI may increase CAM similarity when used to generate standardized text, but may reduce similarity when used to process client-specific information and enhance auditors’ judgments. Using news announcements of generative AI adoption by CPA firms, we find that CAMs become less similar across clients within the same topic following adoption, consistent with more firm-specific disclosure. Using GPTZero to evaluate approximately 4.3 million words in CAMs, we find that AI-detected content was near zero before ChatGPT’s release and increased substantially afterward, suggesting that generative AI became prevalent in CAM writing during the post-ChatGPT period. Finally, we find that following generative AI adoption, auditors shift their efforts toward audit procedures that are less susceptible to AI replacement. Overall, the evidence suggests that generative AI enhances the client-specificity of audit disclosures and leads auditors to reallocate effort toward procedures requiring skills that remain difficult to automate.

嘉宾简介| GUEST BIO

Huimin (Amy) Chen is an Associate Professor at the Manning School of Business, University of Massachusetts Lowell. She obtained her Ph.D. in Accounting from Rensselaer Polytechnic Institute and a Master of Accounting (with an Economics minor) from the University of Nebraska. Before academia, she accumulated extensive professional experience, including three years of audit experience at KPMG China, and holds the Chinese CPA license. She also passed the U.S. Uniform CPA Exam. Her primary research interests are capital-market and supply‑chain‑related financial information.


联系方式

地址:上海市四平路1500号同济大厦A楼21楼 | 电话:021-6598 1341

同济大学 版权所有