The Effects of the Use of Natural Language Processing and Task Complexity on Audit Firm Litigation Exposure

• In Event: Technology and Auditor Liability

Abstract

The increasing prevalence of AI technologies in public accounting can improve audit quality (e.g., EY 2018). For example, Natural Language Processing (NLP) is used to audit complex and simple documents (e.g., Fagella 2020; EY 2017), but can have difficulty analyzing non-standardized human language that can be unique to specific contexts (Godula 2018). We examine the effects of the use of this form of AI and task complexity on jurors’ negligence verdicts. We draw on algorithm aversion, attribution theory, and the culpable control model to develop our hypotheses. Results generally support our expectations such that: (1) jurors return a greater proportion of negligence verdicts when the audit firm uses NLP than human auditors; (2) this effect is more pronounced for complex tasks than simple tasks; and (3) this effect is mediated by perceived causation of the adverse outcome and foreseeability of the misstatement. We offer contributions for theory and practice.

Authors

• Junnan Cui, St. Norbert College

• Jesse C Robertson, University of North Texas