ENHANCING ACADEMIC TUTORING: A CONCEPTUAL FRAMEWORK AND PRELIMINARY EVALUATION OF AN AI PLATFORM LEVERAGING LARGE LANGUAGE MODELS FOR CONTENT GENERATION AND KNOWLEDGE AUGMENTATION

Abstract

Academic tutoring, while highly effective for personalized learning, is significantly constrained by the labor-intensive demands on human resources for content creation, assessment, and tailored student support. Traditional digital learning environments often lack the necessary adaptability and personalization to fully address individual student needs and bridge knowledge gaps effectively. This paper presents a conceptual framework and reports on the preliminary technical evaluation of an Artificial Intelligence (AI) tutoring platform leveraging Large Language Models (LLMs) to address these limitations within academic settings, specifically inspired by challenges observed in university programs like the Business Individual Study Program (BIPS) at the Wroclaw University of Economics and Business (WUEB). The platform integrates distinct toolsets for faculty and students. Faculty tools automate educational content creation (quizzes, assignments with evaluation notes, structured question papers). Student tools enhance knowledge synthesis and interaction through AI-driven features (Chat with Documents/Websites via RAG, Mindmap and Knowledge Graph generation). A preliminary technical evaluation assessed the functional reliability of key generative modules. Results demonstrate promising success rates: Quiz Generation (93%), Assignment Evaluation Note Generation (96%), Structured Question Paper Output (87% structural adherence), Mindmap Markdown Validation (98%), and Knowledge Graph Entity Relationship Identification (93% accuracy with multi-shot prompting). These findings suggest that LLMs offer significant potential for automating manual tasks and enhancing interactive learning, providing a viable path towards more scalable, adaptive, and personalized academic tutoring, subject to further development and rigorous evaluation of pedagogical and ethical implications.