Robots can say “I’m sorry,” but it may not help us forgive them after a mistake. A meta-analysis of 22 studies by University of Michigan PhD candidate Connor Esterwood shows that regaining trust, as with humans, is no easy task. And apologies don’t always cut it. 

“Repairing Trust in Robots?: A Meta-analysis of HRI Trust Repair Studies with A No-Repair Condition” was published in the Proceedings of the 20th ACM/IEEE International Conference on Human-Robot Interaction. The paper is authored by UMSI PhD candidate Connor Esterwood and UMSI professor Lionel Robert. 

The goal of a meta-analysis is to provide a touchstone on a topic and assess where research stands now, what current evidence suggests and what researchers might study in the future. Esterwood’s paper points to a need for thinking about trust repair in a different way. 

“Broadly speaking, trust repair does have an effect and is statistically significant, but the effect is really small,” he says. “So small that it’s not practically meaningful. In other words, while the methods we’re using technically work, they don’t work well enough to be useful. That tells us we need to pivot.” 

As people increase their dependence on robots in the workplace and at home, Esterwood’s research is relevant in identifying future research possibilities in the field of human-robot interaction. The most common use of robots today, he says, are self-driving cars. They’re also typically used in specialized environments like Amazon warehouses and some auto manufacturing plants to help build cars. 

But robots are also being piloted and tested for use in delivering packages and working as crossing guards. Esterwood says within the next 10 to 15 years we’ll see more examples of robots being utilized in everyday environments. 

Thinking about trust repair is crucial, he says, especially because mistakes are “inevitable” and they’re increasingly being tested for routine tasks. 

“I see a robot as essentially an AI with a body,” he says. “The embodiment factor really changes things. The fact that a robot has a physical presence fundamentally shifts how we interact with it. Right now, most robots are in controlled environments like factories, but I think that’s just the beginning.” 

Esterwood began his career at UMSI in 2018 after earning a Master of Science degree in user experience and assessment at the University of Tennessee. His fascination with robots began after working with professor Lionel Robert on a project about autonomous buses.

Throughout his time at UMSI, Esterwood has authored a variety of papers on the relationship between robots and humans. He will be graduating in May 2025 with a PhD in Information. His advisor is Lionel Robert.  

“AIs and robots are physical interfaces,” he says. “ It’s almost like designing a social being, which is an extension of UI/UX design, but instead of designing a screen-based interface, I’m designing a social interface. This drew me into how people communicate and interact, and pulled me into robotics and AI because these machines are social agents.

“Now, I get to figure out how to design their ‘brains’, and that’s just so cool.”

After graduation, Esterwood will be staying in Michigan and starting a job at Wayne State University as an assistant professor at the Mike Ilitch School of Business. 

“I’m excited to continue doing research,” he says. “It’s something I love, and I’m really looking forward to teaching, meeting students and continuing to work on the problems I want to solve.” 

Esterwood’s favorite part of being at UMSI has been the supportive environment and the people he’s worked with who “genuinely care.” His advice for incoming students is to be “willing to fail” and to keep an open mind. 

“The reason we’re doing research is because nobody knows the answers,” he says. “You’re used to being right, but now, you’re never going to be right. You’re just going to be less wrong each time. And that’s a good thing. That’s how we learn, evolve, and create something new.”

Read “Repairing Trust in Robots?: A Meta-analysis of HRI Trust Repair Studies with A No-Repair Condition” by UMSI PhD candidate Connor Esterwood and UMSI professor Lionel Robert in the Proceedings of the 20th ACM/IEEE International Conference on Human-Robot Interaction. See the abstract below: 

As robots become more integrated into various sectors, understanding human–robot interaction (HRI) dynamics, particularly trust repair, is crucial for successful collaboration. For this paper, the authors conducted a meta-analysis of 22 HRI trust repair studies with 3,763 participants to evaluate the effectiveness of strategies for restoring trust after breaches relative to offering no repair. The analysis identified three key findings: (1) strategies are differentially effective, showing limited success in restoring trustworthiness; (2) the overall impact on repairing trust is marginal, with a small effect size; and (3) apologies and explanations are the most effective strategies for trust repair. These insights enrich HRI literature by providing a comprehensive evaluation of trust repair mechanisms, offering valuable guidance for future research and practical improvements in human–robot collaboration. 

RELATED

Learn more about Connor Esterwood by visiting his UMSI profile and personal website. 

Check out UMSI’s PhD in Information program and apply today!