AI-assisted auditing and professional skepticism : Auditors’ perceptions of AI-generated audit evidence

Sammanfattning

Artificial intelligence (AI) is increasingly being integrated into many industries, one of which is the auditing profession. In this context, AI-based tools are used to support data processing, anomaly detection and audit documentation. Although this technology may improve analytical capabilities and efficiency, it also reshapes the evidential environment in which audit evidence is both generated and evaluated. With this development within auditing, it also creates complexity for auditors in exercising professional skepticism toward AI-generated outputs. Hence, this study aims to provide a deeper understanding of auditors’ perceptions of exercising professional skepticism when evaluating AI-generated outputs as part of audit evidence. Through a qualitative research design based on semi-structured interviews with auditors, the collected empirical data was analyzed using thematic analysis. Professional skepticism continues to be perceived as a fundamentally human responsibility in AI-assisted auditing, where AI is understood as a supportive tool rather than a replacement for professional skepticism. At the same time, the integration of AI reshapes how professional skepticism is exercised, requiring auditors to evaluate both client-provided information, AI-generated information and the systems producing the outputs. The exercise of professional skepticism in AI-assisted environments is further influenced by factors such as professional experience, understanding of AI systems and organizational conditions including efficiency demands and time pressure. The study therefore contributes to an understanding of professional skepticism in AI-assisted auditing by conceptualizing it as a form of calibrated trust. This is where AI-generated outputs must be assessed critically, contextually and selectively before being integrated into audit decisions.