CompSci Course Guide Bot

As a collaboration between the Department of Computer Science, the AI Competency Centre, and the Centre for Teaching and Learning, the AI Teaching and Learning project ‘Computer Science Course Guide Chatbot’ intended to create an AI-driven course guide for CompSci students to provide instant, tailored academic guidance. This was driven by student feedback that details essential to their courses, including lecture times, eligibility criteria, and sign-up procedures, were difficult to find due to being spread across multiple areas. This chatbot helps to keep students better informed and to reduce unnecessary stress.

Through development, the bot went through several iterations and platforms, from the original version being built in Microsoft Copilot Studio, then trying it out as a custom GPT in ChatGPT, until settling to build the bot using OpenAI’s API so it could be embedded in Moodle. the department’s virtual learning environment. Technical support provided by the AI Competency Centre with Haseeb Ahmad, Senior Research Software Engineer, and Dominik Lukes, Lead Business Technologist, enabled the project to experiment with this selection of tools, allowing them to settle on a solution.

Once the prototype was ready, testing with staff and students revealed strong engagement with the bot being praised for its convenience and accessible tone. Limitations were found in its small data pool, particularly when it lacked knowledge about historical teaching assignments. An important learning was the need for scope control to avoid the bot hallucinating these answers and instead making sure it could say ‘I don’t know’. Its greatest strength though was that it could be embedded in an existing student environment rather than increasing the amount of places information can be found, made possible by the technical expertise from the AI and ML Competency Centre. This tool is now scheduled for full rollout to Computer Science students in the 25/26 academic year.