LPITutor: an LLM based personalized intelligent tutoring system using RAG and prompt engineering.

Area of Science:

• Artificial Intelligence in Education
• Personalized Learning Technologies
• Natural Language Processing

Background:

• Large Language Models (LLMs) are revolutionizing personalized education.
• Intelligent Tutoring Systems (ITS) require adaptive responses to diverse learning needs.
• Existing systems often lack tailored content for varying skill levels and question complexity.

Purpose of the Study:

• To propose LPITutor, an LLM-based Personalized Intelligent Tutoring System.
• To leverage Retrieval-Augmented Generation (RAG) and prompt engineering for customized educational content.
• To adapt learning materials to individual student skill levels and query complexity.

Main Methods:

• Developed the LPITutor model integrating LLMs with RAG.
• Utilized advanced prompt engineering techniques for response generation.
• Evaluated performance based on accuracy, completeness, clarity, difficulty alignment, coherence, and relevance.

Main Results:

• LPITutor demonstrates effective balancing of response accuracy and clarity.
• The system shows significant alignment with the difficulty level of student queries.
• The model successfully generates customized learning content for diverse learners.

Conclusions:

• LLM-driven ITS, like LPITutor, hold significant potential for transforming education.
• The proposed system offers a viable approach to personalized learning at scale.
• Future work should focus on further adaptation and optimization of AI-driven ITS.