AI tutors to help answer questions in your course

RAG (Retrieval-Augmented Generation) combines two powerful capabilities: 

• Search: It finds relevant information from your course documents (slides, notes, exercises, exams, lecture transcripts, etc.).
• Generation: It crafts a natural, coherent answer using a Large Language Model, based on what it retrieved and what it “knows” from its initial training.

This means students do not get generic AI answers, but responses rooted in the actual materials you provided in class.

• Efficiency: Let the bot handle basic or repeated questions.
• Consistency: Every student gets course-aligned answers.
• Customisable: You decide which documents the chatbot has access to, how it responds, and what topics it can address. It should exclude questions about regulatory and legal aspects. For example, it should refuse to answer questions like “can I get a week extra to turn in my assignment”.
• Feedback Loop: Anonymous logs show what students are struggling with, giving you insights to improve your materials or teaching focus. When deployed in a discussion forum, the questions and answers are shared with everyone.

• Faster help for students.
• Course-specific answers based on provided course material. 
• A judgment-free way to ask questions. In the “chat” version students interact in private. In the “forum” version students can choose to post questions anonymously.
• Can interact in multiple languages, even if your materials are in just one.
• When possible, the chatbot will provide links to the original sources for deeper understanding.

• Even though chatbots use documents from the course when answering questions, they could occasionally provide inaccurate or incomplete answers.
  
  • The safest approach is to delay the response and have a teacher or a teaching assistant verify or edit the answer proposed by the chatbot before making it visible to students. This could help teachers provide faster answers to students. However, this approach prevents the chatbot from giving an immediate response.
  • At the very least, and to raise awareness about the impact of potential errors, clearly communicate to students that the chatbot is not an infallible source and encourage them to cross-check answers with official course materials or ask follow-up questions in class.
• Are teachers’ and students’ personal data and copyrighted material handled in a secure way?
  
  • In a RAG-based system, only excerpts of the documents are sent to an external Large Language Model while composing an answer. We are also looking into the possibility of connecting our chatbots to a locally hosted Open Source Large Language Model to keep all data local.
  • The identity of the students and teachers is always kept anonymous when processing questions.