Presentation
Understanding Human-Agent Teaming: A Mixed-Methods Examination of Perceptions and Interactions
DescriptionWhile quantitative research on trust in HATs is well-established, there is a gap in understanding individuals' subjective experiences during trust violations and repair. This study explores how people perceive and respond to these events, guided by three key questions: (1) How do participants describe emotional and behavioral responses to trust violations and repair? (2) How do personality traits influence reactions? (3) What qualitative themes emerge in the dissolution of trust? Using a mixed-methods approach, participants engaged in a simulated search-and-rescue mission with four autonomous agents, during which, one agent committed two trust violations, followed by repair attempts across five conditions (none, individual agents, or full team). Trust ratings were collected, and participants provided written reflections. Qualitative responses underwent thematic analysis, while survey and behavioral data were examined using regression and ANOVA. Five recurring themes emerged: emotional responses (e.g., frustration), performance changes, blame attribution, loss of trust, and expectations of the violation. Emotional responses were most prevalent, with higher frequency in individual repair conditions. Interestingly, repairs from the violating agent were often not recognized as such by participants. Exploratory analyses showed strong correlations among themes, and regression models suggested that individual differences and thematic responses explained significant variance in trust dissolution.
Event Type
Lecture
TimeTuesday, October 14th3pm - 3:20pm CDT
LocationGrand B
Human AI Robot Teaming (AI)