It is almost a year since we began our collaborative exploration of Generative AI tools towards supporting student learning and reducing instructor workload. One of our partners, Yves Lucet—Professor in Computer Science—has been experimenting with Copilot, Gemini, ChatGPT, and Claude for two of his classes: COSC 222: Data Structures (taught using team-based learning) and COSC 406: Numerical Optimization.
He has found GenAI to be useful in complementing learning activities and improving his efficiency, specifically to:
- Generate a potential solution to a problem handed out in class. Students are given 5 solutions to analyze and critique: 4 of the solutions were generated by their peer and 1 of them was generated by Copilot.
- Given that the Copilot solution is usually pretty good and improves through iterations, in the future, he is thinking to rephrase the prompt to ask Copilot to introduce 1 flaw in the solution for students to find.
- Double check some of the instructor-generated solutions created for in-class practice questions. This speeds up writing solutions and increases confidence that the solutions don’t contain errors.
- Generate sample MCQ questions for student assessments to increase efficiency and variety in question bank.
He cautions that GenAI tools sometimes give plain wrong answers to questions requiring mathematics; thus, it is important to take time to review outputs.
For future exploration, Yves is particularly interested in how GenAI tools can help students improve the quality of peer feedback.
If you have been thinking about experimenting with GenAI in your courses, we welcome the opportunity to engage in that exploration with you. Reach out with your questions and ideas: ctl.ubco@ubc.ca.
Also, begin the new term by joining us at our January 9th Teaching with Generative AI Studio to learn about guidelines, resources, and have conversations around your own ideas.
Subscribe