Study Defines Epistemic AI Literacy for Student-AI Co-Programming

This study proposes a new conceptual framework called Epistemic AI Literacy (EAIL) to analyze how students engage with generative AI, particularly in programming tasks. EAIL views AI literacy as an evolving epistemic phenomenon shaped by dynamic human-AI interactions across various domains. The research applies the AIR framework (epistemic aims, ideals, and reliable processes) to examine how students' goals and methods manifest during AI-supported co-programming. Analyzing a large dataset of student-AI programming dialogues, the study identified key dimensions of epistemic aims, such as mastery orientation, and epistemic processes, including outsourcing, explanation seeking, verification seeking, prompt monitoring, and epistemic justification. The findings indicate a widespread deficit in EAIL, with a vast majority of student interactions exhibiting non-mastery aims and less reliable strategies like simple outsourcing or verification. Only a small percentage demonstrated high epistemic engagement, combining mastery goals with advanced justification strategies.