IMPACT OF AI-BASED AUTO-GRADING SYSTEMS ON STUDENT LEARNING CONSISTENCY IN PROGRAMMING COURSES

DOI: https://doie.org/10.65985/APER.2026805538

Authors:

Mr. Ajay Kumar Gupta

Abstract:

The integration of AI-based auto-grading systems in programming education has significantly transformed assessment practices in higher education. While prior research emphasizes efficiency gains and instructor workload reduction, limited empirical evidence evaluates how auto-grading affects student learning dynamics across time. This study examines whether AI based auto-grading systems improve learning consistency, submission punctuality, and performance stability among undergraduate programming students. Using longitudinal data from 240 students enrolled in two parallel introductory programming course sections, one utilizing AI based auto-grading and the other relying on traditional manual grading, the study analyzes semester-wide grade variance, assignment punctuality, and performance trajectory trends across eight structured programming assignments. Results indicate that students exposed to AI-based auto-grading demonstrate significantly lower grade volatility and higher punctual submission rates. Moreover, longitudinal regression modeling shows smoother upward learning trajectories among students receiving immediate AI feedback. However, excessive resubmission frequency slightly moderates trajectory improvement, suggesting potential optimization behavior rather than reflective learning in some cases. The findings suggest that AI-based auto-grading systems function primarily as feedback stabilization mechanisms that enhance learning consistency rather than merely increasing average grades. The study contributes to understanding AI-enabled assessment systems in computer science education and offers policy insights for balanced AI integration in formative evaluation practices.

PDF