Challenges and opportunities for artificial intelligence in auditing: Evidence from the field

Abstract

• We research AI adoption in auditing by large public accounting firms, with emphasis on its challenges and opportunities. • We find that “simple AI” technologies are used widely in audits while “complex AI” tools are only being developed. • AI adoption challenges are related to transparency / explainability, AI bias, data privacy, and robustness / reliability. • Other AI adoption challenges are fear of auditor overreliance on AI, and the need for AI guidance. • We present ideas for addressing these challenges based on our research and lessons from other fields. In this study we research the adoption of artificial intelligence (AI) in auditing by large public accounting firms, with emphasis on its challenges and opportunities. Some previous studies point to delayed adoption of AI in auditing due to regulations and the need for additional safeguards while others document extensive AI implementation. To address this dissensus, we conducted 22 interviews with experienced audit professionals. We find that “simple AI” technologies such as key data extraction from documents and optical character recognition are used widely in audits while “complex AI” tools are only being developed. We find RPA is used to automate repetitive administrative processes while the use of RPA for audit tasks is not as common. We also find that the main AI adoption challenges are related to transparency and explainability, AI bias, data privacy, robustness and reliability, fear of auditor overreliance on AI, and the need for AI guidance. We present ideas for addressing these challenges based on our research and lessons from other fields.