An AI Tool for Course Selection and Registration

Project Objectives

The goal of this project was to build a fully operational AI driven web tool that can assist Georgetown students in class registration and academic planning. From my personal experience and discussions with my peers, I discovered that many students miss out on taking courses that they are interested in due to their difficulty interacting with the course catalog and the lack of a unified resource for exploring course options. Therefore, I hoped to use large language models as a way to help students seamlessly brainstorm courses and discover different educational paths. The tool I developed allows students to ask questions about different majors, minors, certificates, courses, and professors. It acts as a more natural interface for interacting with the vast Georgetown course catalog, allowing students to search for courses using natural language, rather than the often rigid search mechanisms provided by GU Experience. It also allows students to build out potential schedules before registration and track their degree progress, all using natural language!

Project Steps

1. Learn about OpenAI’s API, the langchain/langgraph agent framework, Retrieval Augmented Generation (RAG), and more…
2. Develop two small prototypes utilizing what I’ve learned:
   • A RAG chatbot that can answer questions about the Georgetown computer science bulletin web page. This system works by breaking down the bulletin into smaller, more digestible chunks, embedding these chunks, and selecting only the most relevant chunks for the agent to use as context when answering a question.
   • A SQL agent that can query a small database of computer science classes. This agent takes a question in natural language and converts it to a SQL query. It uses the output from the query as context to answer the user’s question.
3. Combine these two prototypes into a unified ReAct agent that can call different tools depending on the question asked.
4. Continue to add tools to the agent’s toolkit. These include a rate my professor tool, a transcript transcriber tool, a degree audit tool, and many more…
5. Develop a clean and user friendly front end to chat with the agent.
6. Implement memory: create a user profile that allows the agent to store information about a particular student, including their majors, minors, and courses they’ve completed.
7. Create a scheduling tab that allows the student to draft mock schedules in preparation for registration (similar to how a student would utilize Coursicle).
8. Add a degree progress tab that visually tracks how far along a student is in pursuit of their major or minor.
9. Allow for users to create accounts so that their conversation history and profile information persist across conversational threads.
10. Collect feedback from students and make updates accordingly.

What did the use of Generative AI tools help students do better or differently?

The generative AI tool I’ve developed allows students to navigate academic planning in a more personalized and interactive way. By integrating an LLM with Georgetown’s course offerings, students can seamlessly explore different majors, minors, and career pathways without the limitations of rigid systems like myDegree and GU Experience. This AI-driven system enables students to experiment with customized academic plans, better understand course prerequisites, and make informed decisions about their academic journey. Unlike traditional advising, which is often constrained by time and accessibility, the AI serves as an on-demand resource that enhances, rather than replaces, the role of academic advisors.

Team