摘要
在生成式人工智能迅速发展背景下,审计作为高度依赖专业判断的认知密集型活动,正面临技术赋能与能力跃迁的双重机遇。本文基于认知负荷理论构建“技术嵌入—认知调节—判断优化”分析框架,系统探讨生成式人工智能通过任务重构与智能分流缓释内在负荷、交互赋能与情境导向削减外在负荷,以及知识沉淀与推理增益强化相关负荷,优化审计人员的认知加工路径与判断逻辑,进而提升审计判断质量的理论逻辑。结合实验结论与能力边界分析,提出以任务智能重构、环境智能融合、思维智能引导和安全伦理防护为核心的实践路径。研究有助于拓展智能审计领域的认知心理学分析维度,为生成式人工智能赋能审计判断质量的机制建构与实践落地提供系统化理论支撑与路径启示。
Amid the rapid development of generative artificial intelligence,auditing—a cognition-intensive activity that heavily relies on professional judgment is experiencing both technological empowerment and capability transformation.Drawing on cognitive load theory,this study constructs an analytical framework of"technological embedding-cognitive regulation-judgment optimization"to systematically examine how generative AI enhances audit judgment quality.Specifically,generative AI mitigates intrinsic load through task reconstruction and intelligent task allocation,reduces extraneous load via interactive assistance and context-oriented guidance,and strengthens germane load through knowledge accumulation and reasoning enhancement,thereby optimizing auditors'cognitive processing paths and judgment logic.Based on experimental findings and capability boundary analysis,the study proposes a set of practical pathways focusing on intelligent task reconstruction,environmental intelligence integration,cognitive guidance through intelligent reasoning,and ethical and security safeguards.This study expands the cognitive-psychological analytical dimension of intelligent auditing and provides systematic theoretical support and practical insights for constructing the mechanisms through which generative AI enhances audit judgment quality.
作者
单文涛
王永青
Shan WenTao;Wang YongQing
出处
《审计研究》
北大核心
2026年第1期90-103,共14页
Auditing Research
基金
国家社会科学基金一般项目(项目批准号:23BGL042)和国家社会科学基金青年项目(项目批准号:25CGL021)的资助。
关键词
生成式人工智能
审计判断质量
认知负荷理论
人机协同
智能审计
generative artificial intelligence
audit judgment quality
cognitive load theory
human-AI collaboration
intelligent auditing
