Project members
Jennifer Watson, Edward Crichton, Lucy Sajdler (MPLS Division).
Project summary
Using AI to create a course guide that helps Computer Science students with personalised course selection and faculty expertise, improving academic planning and resource utilisation.
View final project report (PDF)
AI in Teaching and Learning at Oxford Knowledge Exchange Forum, 9 July 2025
Findings from projects supported by the AI Teaching and Learning Exploratory Fund in 2024–25 were presented at the AI in Teaching and Learning at Oxford Knowledge Exchange Forum at Saïd Business School on Wednesday, 9 July 2025.
Project team members each presented a lightning talk to all event participants, and hosted a series of small group discussions.
Follow the links below to view the lightning talk recording and presentation slides for this project.
View lightning talk recording (Panopto) - TO FOLLOW
View presentation slides (PDF)
Project case study
The CS Course Guide Chatbot was developed for the Department of Computer Science to help students navigate a large volume of distributed information about their studies. Students had expressed a preference for accessing this guidance using natural language, so the project aimed to offer a simple, intuitive way to get answers to common course-related questions.
Using OpenAI’s platform (platform.openai.com), I created a custom assistant. To make it easily accessible, I used the embed code provided by openassistantgpt.io and placed it within a text block on students’ Moodle Dashboards during testing. This ensured students didn’t need to visit another website or log in separately, removing barriers to adoption.
The chatbot was designed to answer questions like “When are the Databases lectures?”, “How do I sign up for Graphics practicals?”, and “Am I allowed to take Machine Learning?”. Because data privacy is a key concern, the chatbot was built to avoid using any personal user data. Instead, it prompts users to provide only essential information (eg course of study and year) when needed. To avoid affecting pedagogy directly, the chatbot was not given access to teaching materials, but it could link users to relevant course pages.
Staff and student feedback was very positive. Some students asked questions beyond the original scope, such as how to manage workload, and the chatbot performed surprisingly well. When asked about information it didn’t know, such as who taught a course in previous years, it attempted to guess, sometimes incorrectly. I then updated the chatbot’s knowledge to include this historical information, but this led to confusion and inaccuracies in its responses about the current year’s course details. To address this, I clarified its purpose and restricted its scope. It now responds with a statement along the lines of “I do not have that information” when asked about unsupported topics, rather than attempting to guess.
Testing was most successful when I attended student events and asked participants to try the tool in person, after other volunteer recruitment methods failed. Feedback will be gathered again a few months into full rollout.
The chatbot will be deployed to all Moodle users’ Dashboards from the start of the 2025-26 academic year. Hosting it on local infrastructure is a future goal.
