FICO collaborates with HBCUs to combat data bias 

By Dr. Deborah Bailey
AFRO Contributing Editor 

The Fair Isaac Corporation, or FICO, is one of America’s leading credit reporting agencies. Now, the company has developed a partnership with historically Black colleges and universities (HBCUs) with a goal of getting more Black students to consider a career in data analytics.

FICO uses data sets created using artificial intelligence (AI) to provide a measure of credit risk (the FICO score) for more than 232 million Americans, or 90 percent of the credit eligible population.

For the past two years FICO has sponsored the HBCU Educational Analytics Challenge, which is focused on training students at HBCUs in how to interpret and fine tune AI data sets. This training gives aspiring Black professionals leverage in correcting data bias negatively affecting consumer finance decisions ranging from assigning credit scores and correctly identifying consumer fraud to financing home and car loans. 

According to a 2021 report issued by the Stanford University Human Issued Artificial Intelligence Center, AI data sets disadvantage Black and low-wealth communities with what is referred to as “noisy data”  – or information that can negatively impact credit scoring models by  institutions like FICO and others, when misinterpreted. 

“The way data is interpreted is now a part of our everyday lives,” said Velma Lotson, PhD., assistant professor in Bowie State University’s technology and security department. Lotson is one of the original advisors for FICO’s HBCU Education Analytics Challenge. 

“When aggregate financial data assembled about our communities and neighborhoods is  only interpreted through one lens; it can make a major difference in financial decisions being made about the Black community. If these data sets continue to be used with no cross checks involving Black people and other data professionals of color, the data can cause  long-range negative consequences,”  Lotson said. 

FICO’s Educational Analytics Challenge has included Alabama State University; Bowie State University; Delaware State University; Morehouse College and now, Drake State Community and Technical College; Fayetteville State University and North Carolina Central University.

The Spring 2025 challenge focuses on teaching students to use data responsibly to fight payment transaction fraud. Financial fraud is often committed today by mimicking legitimate banking behavior to secure billions in customer funds, and banks need to leverage transactional analytics to identify and prevent fraud, according to FICO representatives. 

The Federal Trade Commission reports fraud has a disproportionate negative impact on the Black Community. 

Scott Zeldi, PhD., chief analytics officer at FICO is responsible for AI and analytic innovation. He has been the brain power behind the team of FICO professionals that come to campus for meetings to train students on understanding and interpreting data sets. 

“One of the biggest challenges of building predictive models is identifying and handling inherent biases in the data,” said Zeldi. “To help further inclusion by fighting biases in models, we are training and supporting a diverse and talented group of young people to drive the future of AI innovation and propel responsible AI practices forward.” 

As the FICO data challenge grows, students like Bowie State’s computer technology students Bryson Stringer and Shaynsse Alexander both say the experience was an eye opener, reinforcing the importance for data to be analyzed from different perspectives and practical applications for their academic work.  

“What at first glance looks like a fraudulent point of sale transaction to one person due to the location and time of the transaction, may need to be reviewed by at least one other individual,” Stringer said. “I definitely want to continue with interpretation of AI data sets after this experience,” added the Bowie State junior. 

Alexander has seen how important it is to have diversity in the rooms where decisions about fraud and credit scores are being made. 

“One of the fraud models we examined during the FICO challenge actually happened to a family member, who honestly needed to swipe her card several times at a convenience store before it worked,” said Alexander, a first year graduate student with Bowie State’s computer technology program. “It demonstrated to me why someone with my background and experience needs to be in the room when these kinds of data sets are discussed and decisions are made.”