Humanizing AI Grading: Student-Centered Insights on Fairness, Trust, Consistency and Transparency

AI systems lack crucial contextual understanding: Students expressed concerns that AI grading failed to account for individual circumstances and nuanced problem-solving approaches, highlighting that automated systems cannot replicate the contextual judgment human instructors apply when evaluating work.

Human oversight is essential for fairness: The study recommends AI grading function as "supplementary tools under human oversight" rather than autonomous decision-makers, suggesting that responsible AI deployment in education requires meaningful human involvement to ensure equitable treatment of students.

Transparency and explainability build trust: The research, guided by ethical principles including transparency, indicates that students need to understand how and why AI systems make grading decisions—a finding that applies broadly to responsible AI systems beyond education.

Personalization and empathy cannot be automated: The study emphasizes that equitable AI systems must "reflect human judgment, flexibility, and empathy," suggesting that some educational values—particularly those requiring individualized understanding—may be fundamentally difficult for AI to replicate without human collaboration.

Student voices should guide AI design: By centering student perspectives (n=27), this research demonstrates the importance of including affected users in responsible AI development, rather than designing systems based solely on technical capabilities or institutional convenience.