Automated Grading with AI Using Rubrics in Computer Science 1

Abstract

A precise evaluation is essential to provide valuable information that allows for adjustments in teaching and improvements in the learning process. Rubrics are assessment tools that define clear and specific criteria for grading assignments. This study presents a detailed comparison of the evaluation of Programming 1 assignments using rubrics applied manually by different instructors and through Artificial Intelligence (AI) tools on the same assignments. The effectiveness and consistency of the corrections made by AI are analyzed in comparison to those made by instructors and among the instructors themselves. Additionally, surveys conducted with the instructors involved in the grading process are presented to gather their perceptions regarding the accuracy, usefulness, and impact of AI on their work, as well as its comparison with traditional assessment methodologies.