Kalinda Ukanwa: Studying the Algorithmic Biases

More than 20 years ago, Kalinda Ukanwa uncovered algorithmic biases in her work as a data scientist in corporate America. She became a scholar so she could study the issue more deeply

May 22, 2022
Kalinda Ukanwa USC Marshall
Assistant Professor of Marketing Kalinda Ukanwa.

During her 15-year corporate career, Kalinda Ukanwa was steeped in analytics, modeling and data science with the Walt Disney Company, Citigroup, the Transportation Security Administration, Viacom and others.

“I started to see examples of algorithmic bias and fairness in my work,” says Ukanwa, assistant professor of marketing at USC Marshall since 2019. “I thought, ‘Wait a minute. This doesn’t make sense. Why is this kicking out?’”

For example, she observed two people with the same credit score but from different gender or racial groups getting different decisions about lending. “Seeing that from a practitioner standpoint was puzzling,” she says.

While we now hear plenty of references to algorithmic bias and fairness in services, lending and social media, among many other areas, it wasn’t a well-identified concept in the early 2000s.

In her PhD program in quantitative marketing at the University of Maryland, Ukanwa developed a better understanding of the mechanisms behind decision-making models and analytics — and saw an opportunity for research.

“Algorithmic bias and fairness was an important topic to bring to the table with the rise of AI technologies. That’s how I got involved,” she says.

But fairness impacts us all. Ukanwa points out, “No one wants to be treated unfairly. I can’t stress enough how it is in the best interest of everyone to examine these questions now, while it’s relatively early in the development of AI, before things become too entrenched.”

In her PhD program in quantitative marketing at the University of Maryland, Ukanwa developed a better understanding of the mechanisms behind decision-making models and analytics — and saw an opportunity for research.

Ukanwa is currently revising her latest paper on algorithmic bias in service for the Journal of Consumer Research. She looks at the consequences of bias for the firm and consumers. While she finds that prior research is correct, biased algorithms are more profitable than unbiased algorithms, that is only true in the short run. “In the long run, when you take into account the fact that consumers from different demographic groups talk to each other, consumers can gravitate over time to the firm that has the fairer algorithm. The firm with the fair or unbiased outcome can be more profitable over time,” Ukanwa explains.

The majority of firms aren’t even aware that their algorithms can be unfair. “Many of the algorithms that we have today are based on decades of mathematical principles about how you set up optimal models, and that has been built into the current methods of AI,” Ukanwa says.

This area of algorithmic fairness, she says, is relatively new. “There’s been work in the machine learning and computer science community, but even that is very nascent compared to the research on AI in general. And certainly, we’ve barely scratched the surface in the business academic community.”

Ukanwa earned a B.S. in industrial engineering from Stanford (it was the closest match for her aspiration to do analytics in the service of business since the university didn’t offer a BSBA). “If I recall correctly, about 20% of us were women. We banded together, and that played a huge role in me pushing forward and then going on to my master’s at Stanford,” she says.

Later, she returned to Stanford for her MBA with the encouragement of Disney, where she had launched her career as a senior industrial engineer and transitioned to the finance group to focus on business planning and analysis around ideas for new concepts at Walt Disney Attractions Worldwide. “What a great way to start off in a career. I loved that job; it was a lot of fun,” she says. “I really fell in love with solving problems with analytics and modeling. That set the stage for the rest of my career.”  

Ukanwa landed her first executive role at Viacom, where she was director of finance and administration for BET Digital. During that time, she attended a conference that recruited underrepresented minorities for doctoral programs in business. Ukanwa had enjoyed reading journal articles starting as an undergrad, and she often referred to the latest research while solving problems in her professional roles. Her first publication, co-authored with a professor, came about during her MBA program. That conference, she said, helped her realize she really wanted to do academic research. At the time, she was married and pregnant with her daughter, so it didn’t happen right away, but after an executive director position with Kaplan Test Prep, she made the move to academia.

With her cutting-edge research in quantitative marketing, Ukanwa had a lot of opportunities when she went on the academic job market. But Marshall stood out. “For one, the marketing department has a lot of stars,” she says. “I was a fan of many of my colleagues’ work before I came in.”

Teaching students the technical skills they’ll need to thrive in business is important to her as well.

“Big data, analytics, and AI has become central to so many business, non-profit, and governmental operations,” she said. “It is critical for future business leaders to understand how these technologies are being used in decision-making, how to interpret their outputs, and how to communicate it to others. For this reason, it is important that future business leaders learn not only basic data science skills but also the art and science of data storytelling—two skillsets that I emphasize in my Marketing Analytics class.”