Ilitch School professor Connor Esterwood explores the future of trust in AI and human interaction

Placing trust in other humans is at the core of what makes a society successful, but today we also place a lot of trust in machines to accomplish daily tasks. Connor Esterwood, assistant professor of technology, information systems and analytics, has been conducting research into the ways that robot-human trust relationships are built, damaged and repaired.

"A lot of what I do is trust," said Esterwood. "We can trust AIs or intelligent machines too much, or not enough. My goal is to try and manage that or give people the tools to manage that relationship."

Esterwood came to the Mike Ilitch School of Business from the University of Michigan School of Information, where he earned his Ph.D. focusing on human interactions with AI robots and intelligent machines. His interest in this subject began when he was working on designing websites.

"I was thinking about designing better website interfaces. I then moved on to robots, and then I started to move more towards this space where I'm trying to design or create guidelines for better, more reliable, more trustworthy artificial minds," explained Esterwood.

He chose to come to Wayne State to continue his research because he connected with the mission of the university, and its dedication to service and research.

"Service and research collaboration were the two big things that really pulled me in," explained Esterwood. "I visited and everyone was so enthusiastic, and they were engaged with what I was interested in. The energy I get from people at Wayne State is very positive."

Esterwood has wasted no time connecting with others at Wayne State, building bridges between the Ilitch School and the James and Patricia Anderson College of Engineering. This type of collaborative research is very exciting to Virginia Kleist, Dean of the Mike Ilitch School of Business.

"Dr. Esterwood is a strong addition to the excellent research and teaching depth of our Ilitch School faculty. He has been connecting across the university and in the entrepreneurial ecosystem of Detroit to form important research collaborations that greatly elevate the profile of our Ilitch School of Business. I will look forward to Dr. Esterwood's ongoing contributions as he showcases why Wayne State University is a top research institution," said Kleist. "His intellectual interests benefit our students as well, who receive not only the skills they need to begin their careers, but Dr. Esterwood incentivizes a love of learning and a curiosity that will serve them well throughout their lives."

As we interact with more and more intelligent machines on a daily basis, from the algorithms that power social media feeds to the computers inside our vehicles, we are being asked to place trust in these machines.

"Trust is willingness to be vulnerable, and that vulnerability can be based on different beliefs people have in that system," said Esterwood. "That's the relationship I'm looking at now, that trustworthiness assessment that people do when they decide to give their check to a machine and have it cash it all the way to having a stroller go into autonomous mode and drive the baby next to them."

There are three primary factors that influence our trust in machines. Our mental math considers our perception of the machine's ability to do what it says it will do, its integrity or moral consistency and its benevolence toward us and our goals.

"If you have a robot that you want to build a brick wall, if it looks very strong, has a good physical appearance, that can influence the way you perceive the robot's actions," said Esterwood. "There are ways that designers can directly affect perceptions. It takes intentionally designing for it."

Occasionally robots fail. They may make a mistake at their assigned task or communicate with the user in a way that makes a human doubt their benevolence and integrity. It turns out that intelligent machines have a difficult time regaining our trust once it is damaged.

"The most common types of trust-repair strategies that we use in everyday life are short-term verbal strategies. When I mess up, I'll apologize and express remorse or regret that I've offended somebody. Humans do that all the time," said Esterwood. "Apologies really aren't doing much for robots."

Other strategies we use a lot to repair trust include making promises and providing explanations to the affected person. While apologies do not work as well as many think, Esterwood thinks other strategies will work better.

"It seems like something new is needed," explained Esterwood. "Maybe explaining with some re-embodiment. They can say I'm sorry I made that mistake, I'm going to update myself and I'll be better. That's not a promise we would use, but it's a robot specific promise. Exploring that is a new frontier."

Another way to head off a loss of trust would be for the machine to limit the amount of trust placed in it to start. These sorts of trust-mitigation strategies could help users avoid serious consequences that result from over trusting an AI and assigning it tasks outside of its scope.

"I want AI to be trusted in a reliable way, and to have it degrade user trust at times without stopping its use. That's a tricky one," said Esterwood.

As Esterwood works to answer these questions, this technology is still advancing and showing up in nearly every industry and use case.

"It's starting to show up in the wild now, on streets and airports and all these different kinds of places," Esterwood explained. "These machines are emerging and they're becoming useful and productive to people."

That productivity is based on an underlying level of trust, and as we are asked to trust these technologies more and more, they will have to do more to keep, manage and repair our trust. And Esterwood will be there, developing these strategies and leading the way.

-Patrick Bernas, Information Officer III

The Wayne State University Mike Ilitch School of Business prepares students for challenging and rewarding careers, advances the boundaries of scholarly and practitioner knowledge, and enhances the economic vitality of the city of Detroit, the state of Michigan and beyond through its programs, research and community engagement. Established in 1946, the business school was renamed in 2015 in recognition of a $40 million gift from Mike and Marian Ilitch. Thanks to this lead investment, the school moved to a new state-of-the-art building in the heart of the District Detroit in 2018, and academic programming and collaboration with city businesses are expanding. For more information, visit ilitchbusiness.wayne.edu.