Finding the Right Recipe

Cam Kormylo (Photo by Matt Cashore/University of Notre Dame)

As an undergraduate student majoring in IT Management, Cam Kormylo (BBA ‘19) spent plenty of time asking for advice in professors’ offices on the third floor of the University of Notre Dame’s Mendoza College of Business. Now, in his first semester as assistant professor of IT, Analytics, and Operations at Mendoza, his office is down the same hallway.

Kormylo teaches Auditing AI at 8 a.m., which is painfully early for his students but midday for a dad whose two young kids wake him up before dawn. To get everyone moving, he brings a 32-cup coffee maker to each class, where he brews Morning Joe for his students.

“Not all that long ago, I was a student, too,” Kormylo said. “I understand what they are going through.” Back then, he leaned on Mendoza professors Wendy Angst and Corey Angst, Lisa Heming and Laura Glassford in the Office of Undergraduate Studies, and Sondra Champer, who ran Mendoza’s former Cafe Commons.

“Sondra was the most positive person I’ve ever met. Whenever I was stressed, I would sit at the counter and talk to her,” said Kormylo. “That’s what makes Notre Dame so special. Everyone here is just looking for ways to support the students.”

Now, it’s Kormylo’s turn to be the one supporting the next generation of business leaders.

Dungeon Tacos in Dunne Hall

Sometimes, Kormylo still does a double take when he unlocks the office door with his name written on it. As a kid, he never imagined attending a traditional four-year college, much less becoming an academic. In fact, it seemed obvious that he’d go to culinary school and become a chef, and ultimately own his own restaurant.

While classmates in his small town in Wisconsin tuned into Power Rangers, he visited his Italian great-grandmother’s couch to binge-watch celebrity chef Emeril Lagasse prepare delicacies like andouille-crusted redfish. Every year for his birthday, Kormylo’s godparents would drive him into Chicago and treat him to dinner at whatever restaurant was causing the loudest buzz on the famed Chicago Loop’s food scene. When Kormylo turned 14, that was Girl & the Goat, chef Stephanie Izard’s hotspot that permanently disrupted the Windy City’s Restaurant Row by introducing global flavors to fine dining palettes.

“When you’re a kid dreaming of becoming a chef, people think it’s cool, and they are receptive,” Kormylo said. “The waiter took me back to the kitchen, where I met Stephanie. One thing led to another and she invited me to help in the kitchen for a few days.”

Ryan Green, Cam Kormylo and Maddy Kormylo in the Dungeon Tacos kitchen.

That’s how Kormylo started “staging,” working as an apprentice while he was still in high school, helping out and learning. At Girl & the Goat, he was taken under the wing of Jennifer Eisen (now owner of House 406 in Northbrook, IL). “She was the first person outside my immediate family to make me believe in myself,” he said. The experience was so pivotal that it formed the basis of Kormylo’s personal statement when he applied to Notre Dame, which he still calls “my super-reach school.”

“It’s not like I had any crazy academic success to make my application stand out.”

But his personal statement did.

Kormylo wrote about his first night of staging at Girl & the Goat, where he was tasked with preparing 75 soft-shell crabs for diners. “Just snip the genitals,” his essay began. It was “a huge risk” to use such a provocative hook, but one that paid off with an offer of admission.

Imagining himself graduating with a business degree and a plan to open and manage a restaurant of his own, Kormylo went Irish. In his sophomore year, he moved into Dunne Hall, a new dorm with a space for food sales in the basement. It was there that he opened his first restaurant, Dungeon Tacos.

Dungeon Tacos was open three nights a week, had three employees (Kormylo, his high school sweetheart, Maddy, who went to Saint Mary’s, and his friend and roommate Ryan Green (ND ‘18)), and sold three kinds of tacos: steak, chicken and potato.

Sitting at a picnic table outside of Dunne Hall all these years later and gesturing to the basement, Kormylo laughed as he remembered: “We were open from 8 p.m.-2 a.m., and I would start making the salsa at 3 p.m. The three of us powered through, selling 300 tacos a night.” He tried to hire employees, but no one else wanted to work those hours. “I realized that it wasn’t too sustainable.”

It was a realization that led to a tough call.

As much as Kormylo loved being a chef and an entrepreneur, he also wanted to marry Maddy and be a typical dad cheering at their future kids’ soccer games or music recitals. “I realized that I couldn’t be a chef and work until midnight seven days a week and also have the life I wanted,” he said.

By junior year, Kormylo was engaged to Maddy and had declared an IT Management major (now called the Business Analytics major). “I like to pick up things I know nothing about. I couldn’t even turn on a computer and was technically inept,” he admitted.

Trying to reroute his next steps, Kormylo booked an advising appointment with his mentor, Mendoza professor Wendy Angst. “Cam, you need to become a professor,” Angst told him matter of factly.

“I remember nodding my head and thinking, okay. Then that’s literally what I went to do,” Kormylo laughed.

After graduation, he and Maddy married, using their Dungeon Tacos earnings to fund a foodie’s honeymoon to Greece.

After completing his Ph.D. in Business Information Technology at Virginia Tech, he jumped at the chance to “go back home” to Notre Dame as a postdoctoral fellow at the Tech Ethics Lab in 2023.

There, he developed the Auditing AI course, which led to a job offer.

Breaking research on betrayal aversion with AI 

It’s been a big year for Kormylo. In addition to launching his academic career and making the permanent move back to South Bend, he published his first research paper in Management Science: “Till Tech Do Us Part: Betrayal Aversion and its Role in Algorithm Use.” His co-authors are Idris Adjerid of Pamplin College of Business, Sheryl B. Ball of Virginia Tech, and Can Dogan of Radford University.

The paper compares people’s decision-making processes when offered expert advice from human experts versus AI-powered algorithmic experts. Kormylo and his colleagues drew on decades of research in betrayal aversion, which examines how individuals respond when offered advice from an expert (such as a financial advisor).

He said, “If your advisor wrongs you, there will be two costs: you’ll lose money, but there will also be the emotional cost of having been wronged by someone you trusted.”

The paper’s purpose was to investigate how much this effect persists as human experts are replaced by AI algorithmic experts, particularly in financial trading.

Kormylo spent a year designing a financial market simulator game and then ran experiments on Amazon Mechanical Turk, a crowdsourcing marketplace that hires workers for tasks like participating in online surveys. The 300 participants were told they could earn $20-$30 (with some earning as much as $75) by playing 40 rounds of a financial trading game. In each round, they were offered the choice to either make their own decision about how much of their endowment to invest in a risky trade or to defer to a human or algorithmic expert’s advice. They were informed that the “expert” may make a choice out of self-interest that won’t be in the participant’s best interest.

Kormylo expected that “people are still going to feel betrayed by algorithms even if they can’t by definition intend to betray you.”

In reality, the results were the opposite.

When a human advisor helped the individuals and introduced a misalignment, it decreased participants’ future usage of the expert by 16%. However, participants didn’t change their future choices when an algorithmic advisor exhibited the same misstep. Participants continued leaning on the algorithm and ended up earning 20% more than the other group purely by placing their trust in the virtual advisor.

Kormylo said the implications for the financial services industry are clear: While banks and other institutions often assume that anthropomorphizing an algorithm with a face and a name will increase clients’ trust, they may be more forgiving of choices with poor consequences produced by a non-human algorithm.

“There is a phenomenon of organizations trying to hide behind their algorithm’s mistakes. This research doesn’t imply that, ultimately, humans aren’t responsible for their firm’s technology,” Kormylo clarified.

He and colleagues Margaret Traeger, Tim Hubbard, and Mike Villano are conducting similar work to examine how team dynamics shift when working with algorithms that have been assigned male or female voices and names.

Auditing AI

As a teacher, Kormylo aims to teach students how to evaluate and assess AI systems from inside organizations and companies. “We approach the class with the belief that organizations are heavily incentivized to ensure that models they put out are relevant and effectively solve problems,” he explained. “But just as importantly, they have to look at the ethical implications, fairness and data privacy.”

In the first half of the semester, students learn to use tools such as explainability techniques, fairness metrics, interviews and ethnography. Next, they apply their toolkit to case studies. Because AI is rapidly changing, case studies are, too. This is the third time that Kormylo has taught the class, and the case studies have changed each semester to keep up with developments in the field.

“AI is a tool that organizations are going to expect Mendoza alums to use, whether we like it or not. It’s our job to prepare students for the workforce,” Kormylo said. To that end, he has ChatGPT set up in class and shows students how to take advantage of its support, but also what can go wrong if a user isn’t feeding it the right prompts or double-checking its outputs.

And then, after a day at Mendoza, Kormylo comes home and cooks dinner for his family of four. Tacos are often on the menu.