Artificial Intelligence in Automated Assessment and Grading

Abstract

This study looks at the integration of Artificial Intelligence (AI) in automated assessment and grading systems with focus on its role in enhancing objectivity, efficiency, and scalability in assessment in education.  The research design was of mixed methods, which combined the quantitative assessment of the AI-based models to evaluate their performance and qualitative input of instructors and students.  The findings indicated that machine learning algorithms, in particular, ensemble-based algorithms, were over 90 percent accurate in grading as opposed to human raters. This was the case in a broad topic base.  The process also reduced the time on grading by over 70 percent which accelerated the feedback and made students more active.  The statistical studies indicated that there was a high correlation between AI-predicted scores and human ratings (r > 0.85, p < 0.01), and it indicates that the automated framework is trustworthy.  Further, the sentiment analysis of questionnaires showed that majority of students perceived the AI-driven assessment as fair and open, whereas the instructors focused on its ability to help alleviate workload and enhance the process of providing formative feedback.  Despite these strengths, there were issues of making things clear, minimizing bias, and ensuring that they are consistent with teaching objectives. This implied that caution had to be taken in incorporating human control.  Overall, it can be concluded that AI-powered assessment systems can be highly helpful in contemporary education. They are sound, effective and flexible and are compatible with human judgment to ensure that all people are treated equally and that academic integrity is upheld.