Trust and Trustworthiness in Human-Robot Interaction: A Formal Conceptualization
Abstract
The overarching objective of the effort was to explore the possibility of formally characterizing the concept of trust using tools from interdependence and game theory in complex and dynamic social environments. This effort evaluated algorithms for characterizing trust during interactions between a robot and a human and employed strategies for repairing trust during emergency evacuation scenarios. Our results demonstrate that there is a high correlation between our characterizations of trust in a situation and the judgments of people, that timing is a key element necessary for trust repair, and that people tend to overtrust robots, potentially putting themselves in dangerous situations. We have examined human-robot trust in a variety of simulated and live experiments across several different types of risk including both financial and physical risk. These results generally support the conclusion that people will tend to overtrust robots because they believe that the systems are incapable of failure or capable of performing actions or has knowledge which the system cannot perform or does not have.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 11, 2016
- Accession Number
- AD1011180
Entities
People
- Alan R. Wagner
Organizations
- Georgia Tech Research Institute