Student code grew larger; the study cannot say AI caused it
Researchers studied 260 student team projects over ten semesters. Later code was about three times larger and had more ways to handle errors. Comments did not keep pace relative to code size. The team did not track AI use, so this pattern cannot show that AI caused the changes or made students better coders.
What the source reports
The researchers followed code from 260 team capstone projects over ten semesters. Later projects had codebases roughly three times as large as earlier ones. The code also showed greater complexity and more handling for errors. Comments did not keep pace relative to code size. Those are changes in saved code, not direct measures of what each student knew or how each student used AI. The authors did not observe AI use directly in these projects. Time passed alongside the rise of generative AI, but time alone cannot tell us what caused the code to change. A larger codebase does not, by itself, show stronger student skill. The study describes a pattern in team projects and leaves its cause open.
Original source
- Title
- Code, Commits, and Complexity in the GenAI Era: A Longitudinal Study of Team-Based Capstone Projects
- Author
- Asma Shakil, Claus Brabrand, Paul Denny
- Publication
- arXiv cs.CY
- Date
- Friday, November 6, 2026