The Relationship Between Trust and Use Choice in Human-Robot Interaction

Abstract

To understand the influence of trust on use choice in human-robot interaction via experimental investigation. Background: The general assumption that trusting a robot leads to using that robot has been previously identified, often by asking participants to choose between manually completing a task or using an automated aid. Our work further evaluates the relationship between trust and use choice and examines factors impacting choice. Method: An experiment was conducted wherein participants rated a robot on a trust scale, then made decisions about whether to use that robotic agent or a human agent to complete a task. Participants provided explicit reasoning for their choices. Results: While we found statistical support for the "trust leads to use" relationship, qualitative results indicate other factors are important as well. Conclusion: Results indicated that while trust leads to use, use is also heavily influenced by the specific task at hand. Users more often chose a robot for a dangerous task where loss of life is likely, citing safety as their primary concern. Conversely, users chose humans for the mundane warehouse task, mainly citing financial reasons, specifically fear of job and income loss for the human worker. Application: Understanding the factors driving use choice is key to appropriate interaction in the field of human-robot teaming.

How context and design shape human-robot trust and attributions

“…With regard to the influence of individual differences on trust attribution patterns, particularly trust dispositions and attitudes toward robots were positively related to the use intention in two presented contexts. This confirms previous studies on the relation between trust and use intention [13], demonstrating the importance of trustexploring research in rising automation contexts. When it comes to the use of innovative robot technology, future research should investigate not only (dis)trust-related factors but also the extent to which these may be intrinsically motivated and related to people's interpersonal trust perceptions as basis for psychometric analyses (cf.…”

“…Also, especially in fragile usage contexts, as e.g., health care, the idea of being cared for by autonomous robots can, under some circumstances, be incompatible with the requirements of a human and human care which roots in a deep bond between caregiver and patient [12]. In this context, trust is a key factor for the development of users' acceptance toward automated technology and crucial for users' decisions to accept and use or reject a technology [12,13].…”

Can You Trust Your Trust Measure?

“…There is, however, an effort to create more widely-applicable measures [12,29,41], and some of these are beginning to be more frequently used in HRI trust studies. For example, [12], [29], and [21] are seen frequently in HRI studies (for [12] see [7,28,42], for [29] see [4,37,40], for [21], see [16,18,38]). Using the same measures frequently allows for better comparisons across studies, more rigorous and accurate overview of the progress of the field [9], and even comparisons between measures [14].…”

How Internal and External Risks Affect the Relationships Between Trust and Driver Behavior in Automated Driving Systems

“…Examples include user interface design for automotive applications (Miglani, Diels, & Terken, 2016;Noah, Wintersberger, Mirnig, & McCall, 2017); human factors and ergonomics (J. D. Lee & See, 2004;Muir & Moray, 1996;Sanders, Kaplan, Koch, Schwartz, & Hancock, 2019); and human-robot interaction (Chen, Nikolaidis, Soh, Hsu, & Srinivasa, 2018;Freedy, DeVisser, Weltman, & Coeyman, 2007). In this paper, we define ADS trust as the willingness of the driver to be vulnerable to the actions of the ADS.…”