Artificial Intelligence and Auditor Professional Judgement

Artificial intelligence can assist audit analysis while changing how professionals evaluate evidence and accept advice. This systematic literature review examines the evidence supporting concerns about over-reliance on AI in audit decision-making. An OpenAlex title-and-abstract search and supplementary Consensus and SciSpace searches conducted on 29 October 2026 identified 149 records. After deduplication and eligibility assessment, 12 empirical journal studies were included. Five experimental reports, five surveys and two qualitative studies were synthesised using a design-sensitive approach and a six-domain critical appraisal. The experimental evidence more clearly demonstrates context-dependent advice-taking and algorithm aversion than the prevalence of inappropriate acceptance of incorrect AI advice. Survey associations between technology use, trust and reported scepticism do not establish behavioural calibration. Qualitative findings identify responsibility and professional learning as consequential organisational concerns. A proposed framework distinguishes output validity, trust, verification, reliance and defensible judgement, and generates four propositions for future testing. Practical implications include access to contradictory evidence, evaluation of both acceptance and rejection errors, and review that examines reasons for reliance. Limited search coverage, inaccessible reports, version differences, partial full-text access and an AI-assisted singleworkflow process constrain confidence. The review supports proportionate safeguards and better behavioural measurement without estimating a population rate of auditor over-reliance.