Inspection Technology: Observable vs. Unobservable Choice

Abstract In many contexts, a principal, such as a government, relies on inspectors or auditors to monitor companies or individuals to minimize the risks of adverse outcomes, such as environmental hazards or bank failures. In addition to effort, technology has become a crucial factor in this process. Advancements in technology, particularly artificial intelligence, have enhanced the productivity of monitoring efforts. However, rapid technological changes have also increased informational asymmetries regarding the technological capabilities being employed. For example, artificial intelligence washing (the practice of misrepresenting artificial intelligence capabilities to various stakeholders) has become increasingly common and has drawn regulatory scrutiny. This paper examines how the observability of technology influences the inspector's choice of technology in the inspector-inspectee strategic relationship, as well as the associated risk of adverse outcomes, which can be of primary concern to the principal. While prior literature explored cases where technology is unobservable but pre-determined, less is known about situations where both the inspector's choice of technology and effort are unobservable. The results suggest that while higher-capacity technology always reduces the probability of adverse outcomes when technology is observable, the same may not hold ex-ante when technology choice is unobservable. Additionally, when the choice of technology is unobservable, incentives (such as rewards and penalties set by the principal) may be less effective in influencing the inspector's technology choices.