Artificial Intelligence Systems in Accounting and Auditing: A Bibliometric and Exploratory Analysis

Abstract

This study examines the role of artificial intelligence (AI) in accounting and auditing through a two-stage design that combines bibliometric analysis with an exploratory comparative evaluation of AI-based solutions. The current literature shows a fragmentation between conceptual reviews and vendor-driven case descriptions, while a structured cross-cutting view of how AI technologies are actually integrated into financial workflows, and of how this integration differs by entity size and by the trustworthiness profile of the underlying systems, is still limited. The bibliometric component examines 729 peer-reviewed articles indexed in the Web of Science database (post-2015), processed using VOSviewer for keyword co-occurrence analysis. The exploratory component evaluates ten AI-based solutions against a framework that covers AI subset, functional category, target entity size, audit relevance, and disclosed architectural transparency. The bibliometric results indicate that the field is organised around six dominant terms (artificial intelligence, deep learning, machine learning, performance, classification, and model), reflecting a strong methodological convergence on supervised and deep-learning approaches. The exploratory results show that current AI offerings cluster into three functional categories (process automation, analytics and business intelligence, and predictive or audit-oriented systems), with adoption patterns that differ by company size: small and medium-sized entities (SMEs) gain the most benefits from process automation and optical character recognition, while large entities derive higher value from full-population analytics and ensemble-based anomaly detection. The study contributes a replicable methodological pipeline that links bibliometric mapping with applied tool evaluation, a comparative framework that addresses the often undisclosed AI architecture behind vendor black-box tools, and a discussion of trustworthiness limits, including the constitutive (non-error) nature of probabilistic AI failures, with implications for audit assurance and emerging regulatory frameworks such as the EU AI Act and ISO/IEC 42001.