Abstract
Trust has been shown to be an important factor in Human-automation interaction (HAI) because of its presumed effects on behavior. Lack of trust may lead to underutilizing the automation, while over-trust may lead to users relying on automation too much even when it is not perfectly reliable. While this may be intuitively true, reviews of studies have shown that trust has an unclear relationship with actual behavior. While some studies have found a positive relationship between trust and dependence on automation, others have found no such association. Current literature uses various types of tasks to examine trust in automation, from simple tasks (i.e., single-task) like target identification games to more complex tasks (i.e., multitask) such as controlling simulated process control plants where participants have to accomplish multiple tasks simultaneously. This difference in task paradigm, and associated confound in experienced workload, may explain existing discrepancies between trust and automation dependence behaviors. This study examined the idea that workload moderates the relationship between trust and dependence. Results showed that workload significantly affected the relationship. Specifically, when participants experienced low workload, their trust in AI more strongly predicted dependence, whereas this relationship weakened under high workload.
Get full access to this article
View all access options for this article.