Through the FICO Educational Analytics Challenge (FEAC), Scott Zoldi is working to expand opportunities for students at Historically Black Colleges and Universities (HBCUs) by providing hands-on experience in data science and Responsible AI.
The initiative, developed in partnership with the HBCU Data Science Consortium, connects students with real-world analytics projects while pairing them with FICO data scientists as mentors.
“We need to make sure the models everyone builds have proper representation,” Zoldi said. “The only way to do that is to increase the number of students graduating from HBCUs who can contribute to data science.”
Zoldi noted that only about 3% of data scientists are Black, a figure he believes should be significantly higher. Over the past several years, FEAC has reached more than 400 students, exposing them to the ethical and technical challenges of artificial intelligence.
One project challenged students to analyze publicly available housing data from the mid-20th century to identify patterns of bias in lending decisions. While the exercise did not determine the causes of discrimination, it demonstrated how historical data can reveal inequities and highlighted the importance of building AI systems that are transparent and fair.
“It’s not easy to fix bias,” Zoldi said. “But students need to understand these concepts so they can help move AI in a more equitable direction.”
Morehouse College joined the program during its second year and has already produced tangible results. Several Morehouse students have completed internships with FICO, while others have gone on to pursue graduate and doctoral studies in data science.
Zoldi believes exposing students to industry problems helps them envision careers they may not have previously considered.
Beyond education, he stressed the importance of transparency in AI development. Rather than relying on “black box” systems, Zoldi advocates for interpretable AI that allows developers to understand what information models use to make decisions and identify potential sources of bias.
For Zoldi, preparing a more diverse generation of data scientists is essential to building AI that is both innovative and responsible.
