Access to academic support is a major determinant of student success, yet the opportunity to receive help is unevenly distributed. Some students readily approach lecturers or tutors, while others hesitate because of anxiety, fear of judgment, unclear expectations, or low confidence. This gap is pronounced in computing education, where programming tasks accumulate and impose high cognitive load. Although students increasingly turn to general‑purpose generative AI tools, those tools often produce inaccurate, context‑poor, or module‑misaligned answers. To address this, we introduce and evaluate Beacon—a course‑specific Retrieval‑Augmented Generation (RAG) system that offers private, immediate, and module‑aligned academic assistance. Beacon retrieves only approved teaching materials and generates responses grounded in that content, aiming to lower help‑seeking barriers and promote independent learning. Using a design‑based research approach, Beacon was iteratively built and assessed through mixed‑methods evaluation (questionnaires and semi‑structured interviews) with students and staff at a higher‑education institution. Participants reported that Beacon’s answers closely matched course material and were more trustworthy than unrestricted generative AI, especially appreciating the use of pseudocode and scaffolded explanations instead of full solutions. While users remained cautious and would verify AI‑generated answers, they regarded the system as a valuable first point of support before consulting lecturers or official resources. Findings suggest that carefully crafted, course‑specific AI systems can occupy an intermediary space between self‑study and formal support, thereby reducing barriers to academic assistance. Educational AI is most valuable when it broadens access to guidance without replacing the pedagogical role of educators.
Review