Presentation
Quantifying Trust Alignment in Human-Autonomy-Teams: Insights from the Potts Model
DescriptionWe study trust alignment in human-autonomy teams (HATs) using rich naturalistic data from NASA’s HERA missions, capturing full team interaction dynamics over a six-week period.
We introduce an adaptation of the Potts model to quantify how coupling strength between teammates evolves across phases of automation reliability. This approach models trust contagion as a dynamic, socially-driven process grounded in ecological experience.
Our findings show that periods of low reliability amplify trust contagion, with lasting effects on team alignment. We map components of HATs where designers have meaningful degrees of freedom to shape trust dynamics, providing new foundations for engineering resilient human-autonomy systems.
We introduce an adaptation of the Potts model to quantify how coupling strength between teammates evolves across phases of automation reliability. This approach models trust contagion as a dynamic, socially-driven process grounded in ecological experience.
Our findings show that periods of low reliability amplify trust contagion, with lasting effects on team alignment. We map components of HATs where designers have meaningful degrees of freedom to shape trust dynamics, providing new foundations for engineering resilient human-autonomy systems.
Contributors
Event Type
Lecture
TimeTuesday, October 14th3:20pm - 3:40pm CDT
LocationGrand B
Human AI Robot Teaming (AI)