Three strikes and you’re out!: The impact of multiple human-robot trust breaches and repairs on robot reliability
A new study from the University of Michigan found that humans become less tolerant of robots after making too many mistakes, and it’s hard to regain trust.
Like human colleagues, robots can make mistakes that betray human trust. When mistakes occur, humans often perceive robots as unreliable, which ultimately reduces their trust in them.
This study examines four potential strategies for remediating and mitigating the negative effects of these trust violations. These trust strategies were apologies, denials, explanations, and promises of credibility.
An experiment was conducted in which 240 participants worked together with their robot colleagues to accomplish a task, but sometimes the robots made mistakes. provided.
The results showed that none of the repair strategies could completely repair reliability after 3 mistakes.
“By the third breach, the strategies the robot used to fully repair distrust had not materialized,” said Connor Esterwood, a researcher in the UM Department of Informatics and lead author of the study. increase.
Esterwood and co-author Lionel Robert, professor of informatics, say the study also introduces the theory of forgiveness, forgetting, informing and misinforming.
The findings have two implications. Esterwood said researchers need to develop more effective repair strategies to help robots better repair trust after these mistakes. Also, the robot must confirm that it has mastered the new task before attempting to repair human trust in it.
“Otherwise, you risk irreparably losing human trust,” says Esterwood.
What do the findings mean for the repair of human-to-human trust? Trust is never completely repaired by apologies, denials, explanations, or promises, researchers said. .
“Our findings show that trust cannot be fully restored after three breaches and repairs, supporting the adage ‘three strikes and you’re out,'” said Robert. says. “Doing so presents a possible limit to when trust can be fully restored.”
Estherwood said that after a robot makes a mistake, it may not be given the opportunity to do better, even if it can adapt and do better after that mistake. increase.
Lionel pointed out that people can evade or try to evade the robot, causing it to perform poorly. He said this could lead to performance issues, which could result in dismissal for lack of performance or compliance.
Original: Robot: Excuse me. Human: It doesn’t matter anymore!
Than: University of Michigan
