Audit transformation in the era of artificial intelligence: an integrative literature review of AI’s impact on audit quality, professional skepticism, and auditor competence

Purpose – This study examines how artificial intelligence (AI) is reshaping the external audit, focusing on three linked outcomes: audit quality, auditors' professional skepticism, and the competences auditors need. Design/methodology/approach – An integrative literature review of 18 peer-reviewed studies in accounting, auditing and information systems published between 2016 and 2026, complemented by recent standard-setting documents from the IAASB and the PCAOB. Studies are coded by AI application, audit phase, research method and outcome, and synthesised thematically. Findings – AI extends audit coverage from samples to full populations and improves anomaly detection, fraud prediction, the evaluation of estimates and the efficiency of routine procedures. Archival evidence links AI investment by audit firms to fewer restatements, lower fees and more accurate going-concern and internal control opinions. However, the benefits depend on the human–machine relationship: auditors may discount AI evidence (algorithm aversion) or accept it without sufficient challenge (automation bias), and both can impair judgment. Model opacity, data quality, accountability and limited guidance in auditing standards remain unresolved risks, although standard setters have begun to respond. Research limitations – The review is integrative rather than systematic, covers English-language studies only and relies mainly on evidence from the United States, Europe and large audit firms. Evidence on generative AI in auditing is still at an early stage. Implications – Audit firms should treat AI output as evidence to be evaluated, design workflows that prompt challenge and protect the development of junior staff. Standard setters should clarify how much assurance AI-based procedures provide, and educators should combine data analytics with judgment training. Originality – The review integrates technical, behavioural and institutional strands into a single framework in which AI is proposed to affect audit quality mainly through calibrated professional skepticism and auditor competence. It derives five testable propositions and a research agenda relevant to both developed and emerging audit markets.