Samantha Horn: Socioeconomic Differencs in AI Use and Its Consequences For Learning

As generative AI tools become widely accessible, students are increasingly turning to them for help with academic tasks, including developing coding skills. For students from lower-income backgrounds—who often lack access to in-person tutoring or structured support—AI has the potential to serve as a powerful equalizer. But there is growing concern that ease of use may be mistaken for mastery, leading students to develop a shallow understanding of core concepts and to struggle when applying or extending those skills beyond the immediate task.  This project investigates how AI assistance affects students’ ability—and willingness—to develop coding skills in the absence of traditional academic supports. The first study uses a preregistered survey to examine how students from different socioeconomic backgrounds use AI tools for learning, particularly in technical domains. It asks whether students with fewer scaffolds—such as tutoring or peer mentorship—are more likely to rely on AI in place of human support, and how they perceive its effects on their own understanding. The second study experimentally tests whether using AI to complete coding tasks inflates students’ beliefs about their competence, weakens conceptual understanding, or dampens motivation to invest in further skill development. The design also evaluates whether targeted feedback can help recalibrate expectations and improve learning. Together, these studies examine how AI assistance shapes technical skill development and whether it narrows or widens disparities in students’ readiness for the labor market.