基于大模型的关键审计事项研究  

Research On Key Audit Matters Based On Large Models

作　　者：程平[1] 熊俊宇 

机构地区：[1]重庆理工大学会计学院 [2]广州新华学院会计学院

出　　处：《中国注册会计师》2024年第11期24-30,F0002,5,共9页The Chinese Certified Public Accountant

基　　金：国家社会科学基金项目(23CGL074);财政部全国会计重点科研课题(2023KJA3-10)。

摘　　要：关键审计事项(KAMs)在现代审计中占据重要地位。随着审计环境的复杂化,KAMs识别面临许多挑战。在AI时代的背景下,大模型凭借其强大的自然语言处理能力,为KAMs的识别和内容生成提供了新的思路。本文提出了一种基于大模型的KAMs识别与内容生成框架,并对其在提高审计质量、优化审计效率和降低法律风险等方面的应用及整体流程进行了深入探讨。通过构建和微调大模型,本文旨在为注册会计师处理KAMs提供一种基于大模型的技术支持方案。Key audit matters(KAMs)play an important role in modern audit.With the complicated development of audit environment,KAMs identification faces many challenges.In the era of AI,large models relying on the powerful natural language processing capabilities provide new ideas for identification and content generation of KAMs.This article proposes a framework of KAMs identification and content generation based on large models,and discusses its application and overall process in improving audit quality,optimizing audit efficiency and reducing legal risk,so as to provide a technical support scheme based on large models for CPA to deal with KAMs.

关 键 词：大模型 关键审计事项 GLM 

分 类 号：F239.4[经济管理—会计学] TP18[经济管理—国民经济]