Enabling research to improve audit quality

Four insights from the literature on the effects of AI on audit practice

1. AI is improving audit quality and efficiency

The literature consistently shows that, when properly implemented, AI strengthens audit effectiveness. AI helps auditors to identify fraud risks, detect misstatements, analyze entire populations of transactions rather than samples, and reduce human error. What is more, it automates repetitive and time-consuming tasks, allowing audits to be completed more efficiently.

Key message: Overall, AI is enhancing both the effectiveness and efficiency of audit engagements, supporting the profession’s ability to provide timely and reliable assurance.

2. AI changes auditors’ work experience in both positive and negative ways

AI introduces new demands on individual auditors by requiring additional skills and creating uncertainty about future professional roles. These developments may increase mental load and psychological strain. At the same time, AI reduces auditors’ workload and involvement in repetitive tasks, enabling them to focus on more meaningful and value-adding work. The result is a dual effect, where AI simultaneously increases stress while also improving job experience and perceived meaningfulness of auditors’ work.

Key message: Understanding how AI affects auditors’ well-being, engagement, motivation, and job satisfaction is highly relevant.

3. Professional judgment becomes the central challenge

An important concern emerging from the literature is the effect of AI on auditors’ professional judgment and skills. As AI increasingly handles data collection, processing, and basic analytical work, auditors have fewer opportunities to engage in experiential learning, which is essential for building professional judgment and expertise. Reduced exposure to traditional audit tasks may hinder the development of professional expertise and skepticism, particularly among junior auditors. This risk is amplified by the fact that AI can sometimes produce inaccurate outputs that require careful critical evaluation. This creates a challenge: AI imperfections increase the need for critical judgment while potentially weakening some of the mechanisms through which that judgment is developed.

Key message: Future success with AI in auditing will depend not only on technological capabilities but also on maintaining and strengthening auditors’ professional skepticism, independent judgment, and ability to challenge AI-generated outputs.

4. AI should be viewed as an interconnected system of benefits and risks

The literature review demonstrates that AI simultaneously creates resources and demands for audit practice. Improvements in accuracy and efficiency may coexist with deskilling, increased cognitive demands, and challenges to professional judgment. These effects do not operate independently. They influence and potentially offset one another.

Key message: The ultimate impact of AI on auditing will depend not on any single effect, but on how its benefits and risks combine and evolve over time.

Auditors are selected and paid for by the organizations they audit. Policymakers are concerned that this structure influences auditor independence which, in turn, impairs quality. Accordingly, policymakers consider whether audit quality is enhanced if the auditor is appointed by an external (independent) party.

The authors study the working of such a model in a natural setting for local subsidiary audits conducted by the big-4 as part of group audits, where the local auditor is either assigned to the subsidiary through parent firm management or self-selected by the subsidiary.

The authors find that audit partners assigned to subsidiaries receive less information from the auditee, issue fewer going concern opinions, identify fewer control deficiencies, identify and correct fewer misstatements, and are less likely to constrain earnings management compared to self-selected auditors.

Some further preliminary evidence the authors collect suggests that assigned auditors produce lower audit quality through effort reduction.

Audit firms across jurisdictions face a persistent and increasingly acute challenge in attracting and retaining early-career audit talent. A commonly cited explanation for this trend is that today’s students and junior professionals are less willing to accept the demanding working conditions traditionally associated with auditing. While workload and work-life balance undoubtedly play an important role, this explanation implicitly assumes that students possess an accurate understanding of what early-career audit work actually entails.

We argue that this assumption is questionable. Drawing on evidence from our recent Accounting Horizons study (Dierynck, Marangoni, Peters, and Weijers 2025), we suggest that an important, but underappreciated, driver of the audit talent shortage is an expectations-reality gap: a systematic mismatch between what students believe the junior auditor role involves and what junior auditors actually experience in practice. Understanding this gap is critical for audit practice. If students base their career decisions on inaccurate or overly pessimistic beliefs about audit work, firms may lose potential entrants before recruitment efforts can meaningfully engage them. Moreover, misaligned expectations at entry may contribute to early dissatisfaction and turnover, further weakening the talent pipeline.

Audit firms are rapidly integrating Generative AI (GenAI) into their workflows. While these tools can enhance efficiency and support complex judgments, the key challenge is not whether AI provides useful input, but whether auditors use it appropriately. The literature shows that auditors’ reliance on AI is shaped more by behavioral responses, system design, and organizational context than by the underlying technology. Three insights emerge.

First, auditors face a calibration problem. They may under-rely on AI due to algorithm aversion, discounting AI-based evidence, relative to human experts, even when it is equally reliable. At the same time, they may over-rely on AI when outputs appear authoritative, fluent, or easy to use. Both problems impair audit quality: under-reliance biases judgments toward management, while over-reliance reduces professional skepticism.

Second, reliance depends critically on how AI is designed and embedded in the audit process. Features such as perceived control (e.g., the ability to provide input), adaptability of algorithms, and task–technology-fit influence whether auditors trust and use AI outputs. AI is more effective when it aligns with task uncertainty and complexity, and when auditors can meaningfully engage with the system. Poorly designed or poorly communicated tools risk being ignored or misused.

Third, AI affects not only decisions but also how auditors think about decisions. GenAI can improve understanding of complex evidence and help auditors better identify when to raise issues, particularly in remote settings. However, AI can also inflate confidence while reducing self-monitoring, making auditors less aware of when they may be wrong. This creates a risk of overconfidence and inappropriate reliance.

Overall, the literature highlights that successful AI adoption is also a behavioral and organizational challenge, not just a technological one. To realize the benefits of AI, audit firms should consider three key levers. First, governance: providing clear guidance on when and how AI should be used and evaluated. Second, design and communication: ensuring that tools align with task demands and enable auditors to meaningfully engage with the system. Third, training and oversight: developing auditors’ ability to critically assess AI outputs and appropriately calibrate their reliance.

This masterclass explores recent research on the introduction of artificial intelligence in the audit profession and the role of management control practices in enabling its effective use. Based on findings from the Unlocking the Potential of Artificial Intelligence in Auditing project, the session examines how AI tools are being adopted in audit firms, how auditors work with these technologies, and which organizational challenges emerge.

The masterclass is delivered by the research team and provides empirical insights into how AI can be integrated into audit practice in a responsible and effective way.

This masterclass discusses recent research on expanded risk assessment and its implications for audit practice. Drawing on findings from the Auditors at the Crossroads project, the session explores how auditors perceive and apply the expanded risk assessment requirements, and where challenges and opportunities arise in practice.

The masterclass is delivered by the research team behind the project and combines empirical insights with reflection on professional judgment and audit decision-making.