How retrieval-augmented generation and AI agents can make subject-specific LLMs more reliable teaching assistants, giving answers grounded in course material instead of the model's memory.
General-purpose LLMs can hallucinate or drift off the syllabus when used in education. This work combines retrieval over educational content with AI agents to make specialised models more factual and better targeted to their subject.
Presented at the 7th International Conference on Information Technology and Modeling (2024). Published in Technologies of Information and Modeling, Communications in Computer and Information Science (Springer, 2026).