Using Artificial Intelligence to Facilitate Greater Diversity in STEM

According to a Pew Center Research analysis of available data on federal employment and education, as of 2019, African American and Hispanic workers remain underrepresented in the science, technology, engineering, and mathematics (STEM) occupations compared to the total share of the workforce. The same tendency is visible for women in the case of jobs in physical sciences, computing, and engineering. Furthermore, these demographics form only a small share of the STEM graduates.

In January 2022, researchers from the California State University, Fullerton (CSUF), and the University of Southern California (USC) announced that they were exploring the possible application of artificial intelligence (AI) to increase diversity in STEM. The project, titled CareerFair.ai: Increasing Connections to Fast-Growing STEM Careers (CareerFair.ai), strives to spark the interest of a higher number of college students from underrepresented groups and increase their engagement in STEM.

Launched in 2020, it enables students to get their academic and career questions answered via an online platform with the help of AI-based virtual agents. The project’s principal investigators include Yuko Okado, CSUF assistant professor of psychology, Benjamin Nye, USC Institute for Creative Technologies’ director for learning science research, and William Swartout, USC Institute for Creative Technologies chief technology officer. CareerFair.ai’s primary goal is to bridge that gap, and with the support of AI-based technology, to pass on mentors’ advice in a personalized manner to far more students.

Thanks to the platform, which simulates a real-life conversation, the students can communicate with the mentors in real-time. The virtual agents use recorded in advance videos from volunteers: graduate students and real-life STEM mentors. The mentors also come from underrepresented communities, including first-generation college students, women, and racial and ethnic minorities. Professor Okado commented that relatability is crucial because students often look for mentors matching their cultural or personal backgrounds. They must see they are not alone and someone has already trodden the path.

She also noted that first-generation and non-traditional students often work or look after their families. Thus, it can be particularly hard for them to schedule a real-life mentoring appointment. Professor Okado hopes that the platform motivates students to continue pursuing their STEM careers, consider a STEM-related one, or choose a STEM minor. And that it can also help them discover new ways to reach mentors.

CareerFair.ai’s current pilot version comprises five virtual mentors from different backgrounds. The project also works with professional organizations and other institutions that serve minorities, particularly African Americans and Latinx, to expand its mentor network. With the help of a specifically created web platform called MentorStudio, mentors can answer about 260 sample questions for up to 15 hours. The platform also allows them to share advice and personal experience stories.

To ensure the platform matches students’ needs and provides them with mentors from highly-sought fields, the team surveyed 344 CSUF students during the project’s first year. In the fall of the following year, they surveyed 722 more students. Preliminary students’ feedback on their user experience indicated that almost 90 percent of them found using CareerFair.ai a good idea. Slightly over 90 percent of the users also agreed it was clear and easy to understand.

In the spring of 2022, the researchers intend to launch further study encompassing 300 CSUF students to get insights into the platform’s performance related to career advice and information. In the fall of the same year and the spring of the following, they will also seek external feedback.