Effects and Governance of a Hybrid AI–Human Rubric Assessment Model in an Advanced English Course

Abstract

This study investigates the application effects and governance strategies of a rubric assessment method combining artificial intelligence (AI) scoring and human review in the university course "Advanced English." Through an experimental control design, differences between traditional human scoring and AI-assisted scoring were compared, evaluating the role of this hybrid scoring model in scoring consistency, efficiency improvement, and fairness assurance. The results show that AI-assisted rubric assessment significantly improves scoring efficiency while maintaining overall consistency with human scoring results. Scoring consistency saw some improvement, and there were no statistically significant differences across subgroups (e.g., gender), indicating the model did not introduce bias. Student questionnaire feedback indicated that most students held a positive attitude towards this "AI initial scoring + human review" model, believing it enhanced feedback timeliness and scoring fairness. However, a minority of high-achieving students expressed concerns that AI might struggle to fully comprehend complex expressions. Finally, the study proposes governance suggestions to ensure the effective implementation of this model, including refining the rubric, strengthening AI scoring calibration, implementing fairness audits, and providing teacher training. The research indicates that a human–computer combined rubric assessment holds positive potential in Advanced English teaching but requires accompanying governance measures to ensure its reliability and fairness.

Keywords: Advanced English; Rubric Assessment; Artificial Intelligence Scoring; Human Review; Fairness Governance

DOI: 10.7176/IKM/16-1-01

Publication date: January 31^st 2026