Background
Last semester, we developed an AI tutor for FYS2140 Quantum Physics using GPTUiO. The bot was designed around the learning goals and teaching approach of the course, given access to relevant course materials, and successfully used by students throughout the semester. The experience was promising enough that we would now like to see how well the same approach works for other STEM courses.
The goal of the hackathon:
Leave the seminar with a functioning prototype that you can continue developing and potentially use in your course right away.
In this hands-on seminar, participants will build their own course-specific AI tutor. Rather than writing complicated prompts from scratch, we will use one AI as a “tutor-programmer” to help set up a second AI as a tutor. You will describe what you want your tutor to do, and the tutor-programmer will help construct and iteratively improve the instructions for the AI tutor to be used in your course. You will then test the tutor, see where it works well and where it does not, and revise it until it starts behaving the way you want.
To get the most out of the session, please bring:
- A description (up to one page) of what you want your AI tutor to do and how you want it to behave. More detail is better.
- Homework solutions, if you want the tutor to help students with homework.
- A textbook, compendium, or other course materials that UiO has the legal right to store on its servers, if you want the tutor to use these materials when helping students make sense of physics.
- Ideally, a learning assistant or fellow course instructor (your "partner-in-crime”) to help design, test, and critique the bot.
No programming or prior AI expertise is required. The emphasis will instead be on you using your expertise as a physics instructor to decide what kind of help students should receive, what good tutoring looks like, and how you want your AI tutor to behave.
At the end of the semester, we plan to arrange a follow-up seminar where instructors can share experiences, lessons learned, and practical tips from using their AI tutors in their courses.
The workshop will be held by Tor Ole Odden.
Lunch will be served. Please register before September 9.