RAFIKI: Leveraging Large Language Models to Increase AP Computer Science A Enrollment among Disadvantaged High School Females

The gender gap in computing persists even after decades of investment in lowering the gap. Evidence suggests that stereotypical attitudes and bias perceptions play a critical role in limiting female participation in STEM, beginning in middle and high school. The gap is exacerbated in developing nations with limited academic counselor support. Therefore, the goal is to provide early targeted counseling. RAFIKI - "friend" in Swahili is a large language model-based web application designed to mimic an academic coach. Using user inputs, it provides customized academic counseling with curated information on STEM and computing pathways. Initial experimental evidence shows that RAFIKI use leads to a significant increase in AP Computer Science A course enrollment, considered a pathway to future computing career, particularly among female high school students.