Integration of LLMs into Interactive Educational Materials for IT Education

Background and goals

The project aims to explore how large language models (LLMs) can be integrated into educational support tools, specifically to help students learn the fundamentals of IT technology. During the project, we aim to investigate how generative LLMs can be adapted to Bloom's taxonomy to support students at different learning levels. At the same time, we want to examine how Prompt Engineering (PE) can improve communication between students and LLM-supported educational tools.

We aim to integrate LLMs into digital workbooks using the Chain of Thought (CoT) communication model and the Socratic method. These models/methods are used to provide tailored, interactive learning experiences that stimulate analytical thinking and a deeper understanding of new concepts for students, while also providing them with immediate feedback during coding exercises.

There is notable uncertainty regarding how to communicate predictably with generative AI to ensure the generation of reliable information. Several methods have been developed to address this, including Retrieval-Augmented Generation (RAG) and the LLM-Agent Collaboration Framework with Agent Team Optimization.

Objectives and benefits

Educational workbooks for IT training require a clear structure and reliable interaction to help inexperienced students acquire the knowledge and skills outlined in the curriculum. Currently, there are no known systems that provide this functionality, and the race to develop them is progressing rapidly. Such tools are believed to offer higher educational outcomes at a lower cost, making their development a high priority.

We plan to complete a proof-of-concept prototype of an interactive workbook for an introductory programming course at Arcada and evaluate it with students.

Sustainable development goals