Hossain, S. M. A., Shayoni, R. K., Mridha, M. F., & Shin, J. (2026) — arXiv:2607.15738 (cs.CY). 📄 Full text (arXiv)
EduGuard is a retrieval-augmented generation (RAG) tutoring framework that directly confronts the safety and pedagogical failures of unrestricted LLM tutors in introductory programming. Unrestricted tutors hallucinate, contradict course policy, reveal complete solutions, and foster passive dependence; EduGuard counters these with query understanding, instructor-approved course retrieval, pedagogical-strategy selection, rubric-aware generation, claim-level verification, and explicit overreliance control. The design maps onto the ai-tutor-safety-harms taxonomy and directly targets hallucination-risk and over-reliance in deployed tutors.
To make evaluation provenance explicit the authors build BILearn-CS, a 600-query instructor-authored, TA-validated benchmark spanning concept questions, debugging, misconceptions, assignment support, code-mixed Bangla-English queries, and adversarial direct-answer prompts, then validate on a public CS50-style forum set and a small n=10 pre/post-test pilot. This contributes a rare benchmark for safe tutoring and extends the programming-its and llm-feedback-programming-classroom literature with a verifier-separated architecture (DeBERTa-v3-large-MNLI as an architecturally separate claim verifier) that improves on structured-llm-feedback-programming baselines and GPT-4o-mini and Llama Socratic tutors. The work is also a concrete instance of pedagogical-llm-training where safety and pedagogy are baked into the generation pipeline rather than retrofitted.
Related Pages
- ai-tutor-safety-harms — EduGuard's verification + overreliance-control modules operationalise the safety-harms taxonomy for programming tutors.
- over-reliance — Built-in overreliance control prevents passive dependence on the LLM tutor.
- llm-feedback-programming-classroom — A RAG programming tutor evaluated on CS50-style forum queries and a classroom pilot.
- structured-llm-feedback-programming — Outperforms rubric-prompting and same-model RAG baselines via claim-level verification.
- pedagogical-llm-training — Bakes safety and pedagogical strategy selection into the generation pipeline.
- programming-its — Extends intelligent tutoring for programming with instructor-approved retrieval and verification.
- didactical-teacher-assistant-dimensional-modeling — Encoding tutoring strategy in an explicit didactic layer (rather than implicit LLM prompting) makes