Accountability in an Era of Agentic GenAI Coding: Rethinking the Early University Computing Experience

Kathi Fisler, Shriram Krishnamurthi, Michael Littman

ACM Technical Symposium on Computer Science Education, 2027

Abstract

GenAI coding agents now enable people to create software artifacts that are significantly beyond their own manual coding skills. This becomes a cause for significant concern in artifacts with real-world impacts. As educators, we must help students who leverage agents for exploration and productivity to learn to be accountable for their work. What does this look like?

This paper describes our experience from an agentic coding course aimed at 2nd–4th semester university students with at least one semester of programming. We gave students projects that were just at or beyond what we expected they could have written on their own. Students were expected to use agents heavily, but also to defend the quality of what they produced. After each project stage, we taught them software-engineering practices that could have been useful and expected them to use these on subsequent assignments. Students kept reflection journals throughout each project. Students clustered into four broad strategies for taking accountability for their work. We describe these and other observations about teaching early-stage university CS students to program collaboratively with agents.

Paper

PDF


These papers may differ in formatting from the versions that appear in print. They are made available only to support the rapid dissemination of results; the printed versions, not these, should be considered definitive. The copyrights belong to their respective owners.