Impact of permitting Microsoft Copilot use under the course policy

Determine the impact of permitting Microsoft Copilot use under the stated course policy in the three studied upper-level proof-based undergraduate mathematics courses.

Background

The courses explicitly allowed generative AI (Microsoft Copilot) use, with requirements to verify outputs and reference textbook results by number. The instructor observed differences in submissions and student behaviors but emphasized that the true impact of this policy could not be established from the available data.

This unresolved question concerns how allowing Copilot, within the described constraints, affects student work and the course overall, including potential benefits (e.g., awareness of AI limitations) and drawbacks (e.g., quality issues, unusual mistakes) that were anecdotally observed.

References

It is impossible to fully determine the impact of the allowed use in the course but overall, it seemed to raise some awareness of the need for caution in accepting answers generated by AI.

— Gen AI in Proof-based Math Courses: A Pilot Study  (2509.13570 - Klawa et al., 16 Sep 2025) in Instructor Perspective

Finally, it is possible that participants could be less engaged in this experiment than they would be in a graded activity. Future research could investigate the impact of incentives in these interactions by integrating the experiment structure into a course or otherwise incentivise participants perhaps by informing participants their compensation would be tied to performance.

— Your Programming Students' Cognition with ChatGPT: Higher Performance, Lower Retention, and Reduced Ownership  (2609.21194 - Bergh et al., 18 Sep 2026) in Section 5, Limitations and Future Work