L'émergence de l'intelligence artificielle dans le métier de l'audit externe: outil d'optimisation ou menace? Étude des enjeux éthiques, réglementaires et organisationnels.

Abstract

This thesis explores the integration of Artificial Intelligence (AI) in external auditing, assessing whether AI serves primarily as an optimization tool or whether it poses significant risks to the profession. The research focuses on the ethical, regulatory, and organizational challenges introduced by AI in auditing, aiming to understand its broader impact on audit quality, efficiency, and the overall audit landscape. The methodology of the research is divided into two parts: a comprehensive literature review and an empirical analysis. The literature review establishes the theoretical framework by examining the history, definitions, and applications of AI within the auditing profession. It also addresses the ethical, regulatory, and organizational challenges that AI presents. Following this, the empirical analysis is conducted through qualitative interviews with external auditing professionals. These interviews provide insights into auditors' perceptions and experiences, particularly regarding ethical dilemmas, regulatory impacts, and the necessary adaptations within organizations. This research findings indicate that AI has considerable potential to enhance audit quality and efficiency. AI can automate routine tasks, improve data analysis, and facilitate the early detection of risks and anomalies, all of which contribute to more accurate and reliable audits. Additionally, AI allows auditors to process large volumes of data quicker, allowing them to focus on complex, judgment-intensive aspects of the audit process. However, the research also identifies significant challenges associated with AI integration. Ethical concerns, such as algorithmic bias, data privacy issues, and the potential erosion of human judgment, are major challenges. These issues questioned the need for robust ethical guidelines development to govern AI’s use in auditing. Furthermore, current regulatory frameworks may not be fully equipped to address the complexities introduced by AI, highlighting the need for updates to existing standards to ensure they remain relevant and effective. Organizationally, the adoption of AI requires substantial changes, including ongoing training and development for auditors, potential restructuring of audit teams, and managing the risks of job displacement. The thesis concludes that AI can indeed serve as a powerful tool for optimizing auditing processes, but its successful implementation depends on how well the associated challenges are managed. To this end, the thesis recommends developing clear ethical guidelines to address algorithmic bias, data privacy, and the role of human judgment. It also advocates for updating regulatory frameworks to better accommodate AI, investing in continuous training for auditors, and preparing organizations for the structural changes that AI integration might induce.