Teaching Programming in the Age of ChatGPT

In a recent study conducted by Sam Lau from UC San Diego and his advisor Philip Guo, over 20 programming instructors from universities across the globe were interviewed to understand how they plan to adapt their courses as more students have access to AI coding assistance tools like ChatGPT and GitHub Copilot. Short-term plans focus on preventing cheating, as instructors worry that students will rely too heavily on these tools and not learn the material deeply. Strategies such as in-class exams and exposing students to the limitations of AI tools were suggested. For the long-term, ideas were split between resisting AI tools and embracing them. Those who resist fear that students won’t learn the fundamentals of programming, while those who embrace believe it will better prepare students for future jobs. Custom-built libraries and culturally diverse assignments were suggested to make homework assignments AI-proof, while in-person exams and oral exams were considered for assessments. Instructors also saw the potential for AI tools to provide personalized help to students and generate practice problems. Ultimately, the study findings highlight the need for further research and discussions on how to effectively and ethically integrate these tools into computing education.


To top