AI-Enabled Banking and Audit Quality: A Conceptual Framework Linking Explainability, Governance, and Professional Judgment

Abstract

The growing reliance on artificial intelligence (AI) in key banking operations such as credit scoring, anti-money-laundering (AML) monitoring, fraud detection, and automated compliance is changing the information and decision-making context in which financial statement and internal control audits are performed. The literature that addresses AI application in banking has not yet discussed the connection between AI governance decisions in banks and the auditors' professional judgement in the context of the impact on audit quality, and vice versa, the literature related to the use of explainable AI (XAI) techniques in auditing does not consider the impact of the bank's AI governance decisions. In this paper, we propose a conceptual model where AI-enabled banking is an antecedent condition whose impact on audit quality is mediated by two mechanisms, on the bank side (AI explainability and AI governance/accountability) and on the audit side (auditors' professional judgment, and their scepticism), while regulatory regime maturity serves as a boundary condition. The paper presents an agency theory and socio-technical systems theory perspective that suggests that audit quality is not a product of technology but an emergent property of the design of explainability and accountability structures around algorithmic decision systems in AI-intensive banking contexts. Six theoretically-driven propositions are formulated to inform possible empirical tests. The paper enriches the accounting, auditing and banking-technology literatures by providing a model that is integrated and previously fractured into a number of separate research conversations; and it provides implications for banks, internal and external auditors, audit committees and regulators. It is a conceptual study in the sense that no primary and secondary data are analyzed and the model and propositions are just for future empirical research.

Keywords

artificial intelligence; audit quality; explainable AI; AI governance; professional judgment; banking; auditing