Architecture and Usage Information for a Course-Aligned, GPT-Powered Chatbot

Large Language Models (LLMs) offer impressive natural language understanding abilities and are suitable for the development of educational chatbots for engineering students. We present the architectural design of the Oliver chatbot, an LLM application for answering natural language queries over a corpus of course-aligned documents. Oliver is based on a Retrieval Augmented Generation system and is equipped with a user-friendly web interface. We present Oliver’s use-cases and demonstrate how logging data can be used for generating reports on student learning throughout a term. We support this using data obtained during the Fall of 2023, in which Oliver was deployed to a class of approximately 100 Mechatronics Engineering students.