COMPLEX ESTIMATES AND AUDITOR RELIANCE ON ARTIFICIAL INTELLIGENCE

Abstract

Audit firms are investing millions of dollars to develop artificial intelligence (AI) systems that will help auditors execute challenging tasks (e.g., evaluating complex estimates). Audit firms assume AI will enhance audit quality. However, a growing body of research documents “algorithm aversion” – the tendency for individuals to discount computer-based advice more heavily than human advice, although the advice is identical otherwise. Auditor susceptibility to algorithm aversion could prove costly for the profession and financial statements users. Accordingly, we examine how algorithm aversion manifests in auditor decisions using an experiment that manipulates the source of contradictory audit evidence (human specialist versus AI specialist system) and the degree of structure within the client’s estimation process (higher versus lower) for a complex estimate. Consistent with theory, we find evidence that algorithm aversion amplifies the persuasive effect of greater estimation structure, making auditors more likely to discount contradictory audit evidence and accept management’s preferred estimates.