Organizational Behaviour Seminars

Despite the growing availability of algorithm-augmented work, algorithm aversion is prevalent among employees, hindering successful implementations of powerful Artificial Intelligence (AI) aids (AI-based decision-support systems). Here, we examined the effects of two distinct aspects of such systems – Providing explanations to the user versus Requesting explanations from them – and their interplay with decision makers’ uncertainty avoidance (UA) in shaping their supportive attitudes toward such systems. A pilot study among U.S. employees revealed that both UA and the feature of providing explanations to the user (but not the feature of requesting explanations from them) were positively associated with employees’ supportive attitudes toward a recalled AI aid they currently use at work. Importantly, two preregistered experiments, manipulating both the system’s explanations features, resulted in causal evidence for the positive effects of (1) deploying a system that provides explanations, on all employees, but in particular, on high-UA employees, (2) a distinct preference, especially by high-UA employees, for complete )versus partial( explanations provided by the system; and (3) an enhanced preference by high-UA employees, for systems that request users to provide explanations that will be used to train the system rather than to document their decision-making process. Our studies provide insights into improving the management of Human–AI collaboration by integrating knowledge about system features, their optimal presentation to employees, and the characteristics of employees best suited to work with such systems. As such, they also contribute to understanding, mitigating, and managing employee aversion to powerful AI aids, as well as informing their effective design.

TBA

TBA

TBA

TBA

TBA

TBA