Educational Application of RAG-based LLMs

The integration of Large Language Models in the education field has highlighted issues like spreading of misinformation or wrong data, misalignment with curriculum or syllabus, and lack of personalized learning experiences. So, this research proposes a web-based intelligent tutoring system which combines concepts of fine-tuned Large Language Models and Retrieval Augmented Generation. This will help provide precise, personalized and curriculum-aligned assistance in studying. The proposed system will be able to process documents of different formats and then, transform and store them into a vector database for proper and semantic retrieval, and will cite or refer verified course materials in it’s responses using parameter-efficient fine-tuning. This ensures that system is reliable and adaptable with smooth conversations. Initial testing and analysis show improvements in accuracy, relevance and student satisfaction as compared to conventional Large Language Models. It will also be secured using OAuth, JSON Web Tokens, or other login security methods. The proposed system provides a cost-effective and scalable basis or template for creating domain-specific Artificial Intelligence tutoring systems, which align with objectives of education.