Improving Coding Assignments with Partial Auto-Grading and Immediate Feedback: Report from an intervention

Paper i proceeding, 2026

This paper presents a case report on the use of partial auto-grading with immediate feedback in the course Introduction to Artificial Intelligence (MMS131) at Chalmers University of Technology. The intervention aimed to reduce grading workload while providing students with timely, formative feedback on coding assignments. Evaluation was carried out by comparing the 2023 course iteration (manual grading) with the 2024 iteration (auto-grading integrated into Jupyter notebooks). The case showed reduced grading time, shorter feedback delays, and lower resubmission rates, alongside
generally positive perceptions from both students and teaching staff. Challenges included installation issues and occasional concerns about the clarity of automated tests.We discuss the pedagogical and operational implications of these findings and reflect on design choices such as scaffolding with a preliminary assignment and balancing visible and hidden tests.