ARTIFICIAL INTELLIGENCE IN AUDITING: EVALUATING THE IMPACT OF GENERATIVE AND PREDICTIVE MODELS ON AUDIT QUALITY AND EFFICIENCY

Abstract

Artificial intelligence (AI) is significantly contributing to beneficial transformations in auditing by enhancing analytical capabilities, automating monotonous tasks, and refining decision-making processes. This piece of research looks into the effects of predictive and generative AI on the quality of auditing as well as its efficiency. Moreover, the research examines how professional skepticism can influence the auditing process in environments where AI is actively used. The research was conducted on the basis of an applied explanatory design with the use of a quantitative cross-sectional method. Data for the study were gathered from a sample of 214 general auditors who work for either the Big Four or other leading international firms, together with information regarding the performance of 38 audit engagements. The application of hierarchical regression and Structural Equation Modeling (SEM) allowed researchers to prove their hypotheses. The current research indicates that predictive AI is able to positively influence the quality of auditing (fraud detection, anomaly detection, risk assessment, and so on) (β=0.53, p<0.001). Generative AI could improve the efficiency of auditing due to its key functions that include automation of documents, reports, and financial data (β=0.47, p<0.001). Additionally, professional skepticism could improve the quality of auditing as well (β=0.29, p<0.01). The results of the multiple regression model accounted for 52% of the overall audit performance, while the results of SEM (χ²/df=2.11, CFI=0.94, RMSEA=0.061, SRMR=0.054) showed satisfactory goodness of fit of the model. The findings of the study make a significant contribution to academia, providing the empirical basis for highlighting the critical roles of predictive and generative AI technologies. Moreover, the research findings are of great practical value for audit companies that are planning to implement AI into auditing processes. Finally, the results show that there is a necessity to create frameworks for AI-led auditing processes that should strike a balance between technical advancement and human skills.