The Role of Environmental Predictability and Costs in Relying on Automation

TLDR: It was found that participants underutilized automated advice in more predictable environments and when advice was more costly; however, when costs were low and the environment was less predictable, participants tended to overutilize automated advice.

Abstract: There is a growing need to understand how automated decision aids are implemented and relied upon by users. Past research has focused on factors associated with the user and automation technology to explain reliance. The purpose of the present study was determining how the predictability of the environment affects reliance. In this paper, we present the results from an experiment using a digital game where participants had access to a free environmental cue of varying predictive validity. Some participants also had access to automated advice at varying costs. We found that participants underutilized automated advice in more predictable environments and when advice was more costly; however, when costs were low and the environment was less predictable, participants tended to overutilize automated advice. These findings provide insights for a more complete model of automation use, and offer a framework for understanding automation biases by considering how automation use compares to a model of optimality.