ISCAP Proceedings: Abstract Presentation
Framework for Proper Use of Generative AI in Computer Science Classes
Teko Jan Bekkering
Northeastern State University
Abstract
The advent of ChatGPT has significantly transformed Computer Science education. We have observed both opportunities and challenges that this AI tool brings to the classroom. One of the most notable influences is its ability to enhance student engagement and learning. As an interactive tool it can provide instant feedback and support to students. For instance, when students have difficulty understanding programming concepts, they can query the AI for explanations and examples. Using the AI to generate code snippets and debug errors allows students to experiment and learn through trial and error without waiting for instructor support. On the other hand, critical thinking and problem solving can suffer if students become too reliant on this tool. Vendors are now targeting education for use of their LLMs (Google, 2024; Microsoft Education Team, 2023; OpenAI, 2024) and try to provide educators with support for inclusion in their courses (OpenAI, 2023).
There is no doubt that students, including CS students, use generative AI for their academic work. The challenge is to teach them how to use it responsible and effectively. Firth at al. (2023) presented a framework of two matrices for student decision support, with the first matrix representing the type of work and the second matrix how it would be used. The matrix framework was a general framework, not targeted to education but general use. Later, Firth et al. (2024) revised their framework specific to education using a single matrix with students on one axis, and professors on the other.
Universities start to provide guidance on the use of AI for coursework, but different majors have different needs and individual faculty may have their own instructions, wishes and concerns. Our current work focuses on the development of a decision-making framework for CS courses taught by individual faculty. Decisions are made for the educational tasks in those courses and consists of two matrices. The first matrix concerns the nature of the work with depth of knowledge and principles vs. developing programs on the two axes. The second matrix concerns the completeness of the product and use for practice vs. submission for credit on the two axes. As in the original work of Firth et al. (2023), quadrants are color coded as green, yellow, and red. Using the two colors from the matrices, faculty can provide easy-to-understand guidelines for the appropriate use of generative AI in their CS courses. With minor modifications majors, the framework may also be adaptable for other academic majors.
References:
Firth, D., Derendinger, M., & Triche, J. (2023). Cheating Better with ChatGPT: A Framework for Teaching Students When to Use ChatGPT and other Generative AI Bots. Proceedings from the ISCAP Conference, 9. https://doi.org/10.62273/BZSU7160
Firth, D., Derendinger, M., & Triche, J. (2024). Cheating Better with ChatGPT: A Framework for Teaching Students When to Use ChatGPT and other Generative AI Bots. ISEDJ, 22(3), 47.
Google. (2024, May 16). Bringing Gemini to Google Workspace for Education. Google. https://blog.google/products/classroom/google-ai-gemini-workspace-for-education/
Microsoft Education Team. (2023, December 14). Expanding Microsoft Copilot access in education. Microsoft Education Blog. https://www.microsoft.com/en-us/education/blog/2023/12/expanding-microsoft-copilot-access-in-education/
OpenAI. (2023, August 31). Teaching with AI. https://openai.com/index/teaching-with-ai/
OpenAI. (2024, May 30). Introducing ChatGPT Edu. https://openai.com/index/introducing-chatgpt-edu/