The level of artificial intelligence application and audit efficiency

The integration of Artificial Intelligence (AI) into business operations is reshaping auditing by changing data structures, business processes, and control environments. Although AI is often expected to shorten audit cycles and improve audit efficiency, it may also create additional audit work because of model opacity, extended business chains, and emerging technical risks. Using Chinese A-share listed firms from 2015 to 2024, this study examines the association between Artificial Intelligence Application Level (AIAL) and audit efficiency based on 4,931 firms and 36,582 firm-year observations. Audit efficiency is measured primarily by audit report lag, defined as the time between the fiscal year-end and the audit report signing date. Firm-level AIAL is constructed from annual report disclosures by counting the frequency of AI-related terms, including ‘artificial intelligence,’ ‘AI,’ ‘machine learning,’ ‘deep learning,’ and ‘intelligent algorithms,’ followed by a logarithmic transformation. Financial, governance, and audit-related data are mainly obtained from the CSMAR database, while the AIAL indicator is derived from listed firms’ annual report texts. The empirical analysis employs firm fixed effects and year fixed effects models and further examines whether the AIAL–audit efficiency association varies with auditor professional ability, management internationalisation, and management's technological governance capability. The results show that higher AIAL is associated with longer audit report issuance time, suggesting that AI-related disclosure intensity is not necessarily associated with higher audit efficiency. Instead, it may reflect greater business and technical complexity faced by auditors. This positive association is significantly weakened when firms are audited by Big Four auditors and when executive teams have stronger IT backgrounds, indicating that auditor professional ability and management technological governance capability help reduce the audit burden associated with AI-related complexity. The moderating role of management internationalisation is less consistent. Robustness tests using an alternative audit efficiency measure, lagged AIAL, and industry fixed effects provide consistent evidence. This study contributes to the literature by shifting attention from AI as an audit tool to AIAL as part of the audited firm's technological environment. It also shows that auditor capability and firm-level technological governance are important complementary conditions in AI-intensive audit settings. Because AIAL is constructed from corporate disclosures, the findings should be interpreted as evidence on AI-related disclosure intensity rather than as a direct measure of the actual depth of AI implementation. Overall, the study highlights the need to align AI-related business transformation with audit capability and managerial technological governance.