"Can I get a second opinion?" How user characteristics impact trust in automation in a medical screening task

Abstract

As technology advances, processes traditionally carried out by humans are being automated in a variety of industries, such as automotive, security, and food service. In the medical field, advances in automation allow for disease classification, diagnosis, and even treatment recommendations. Technological advances have improved diagnoses by automated devices such that many cases can be more accurately diagnosed by a computer program than by a medical doctor. The hindrance to implementing these technologies is that these systems need not only to exist, they must be accepted, trusted, and appropriately used by both patients and healthcare providers. Previous literature on automation acceptance has focused primarily on how design features and characteristics of the automation influence human trust. Less research has explored the role that user characteristics—such as personality and dispositional traits—play in developing trust. User responses to automation may warrant adaptation in how automation is presented and distributed in order to encourage its acceptance. In the present study, researchers examined the relationship between user characteristics, trust, and automation use in a medical screening decision task. Although user characteristics were found to predict trust attitudes, they did not significantly predict trust behaviors, i.e., automation use. These findings are discussed with the consideration of the differences between attitudes and behaviors in predicting trust. Keywords: trust in automation, medical decision-making, trust, automation, user traits