Toward Fair and Ethical Teacher Evaluation: A Qualitative SLR on AI-Enhanced Value-Added Approaches

Abstract

Abstract
The rapid integration of artificial intelligence (AI) into education has reshaped how student learning is monitored and how teachers are evaluated, raising new questions about fairness, ethics, and equity. This qualitative systematic literature review synthesizes findings from 24 empirical and conceptual studies to examine how Value-Added Models (VAM), knowledge tracing techniques, and generative AI (GenAI) tools are being integrated into teacher evaluation systems. The review identifies three dominant themes: (1) persistent validity and bias concerns in traditional VAMs, particularly regarding socioeconomic status (SES), cultural differences, and missing data (Chetty et al., 2014; Rothstein, 2010); (2) the emerging role of machine learning and autoregressive knowledge tracing models in improving prediction accuracy and capturing students’ self-regulated learning behaviors (Piech et al., 2015; Zhou et al., 2025); and (3) the ethical implications and equity risks associated with AI-enhanced assessments, including AI trust, cultural fairness, and teacher anxiety (Kong & Yang, 2024; Tossell et al., 2024). The synthesis highlights that while AI-enhanced VAM frameworks show promising improvements, with reported accuracy gains of 5–12% and bias reductions to under 5%, there is limited qualitative evidence on how teachers, students, and school leaders perceive these systems. The review proposes a human-centered, ethically informed framework that integrates GenAI transparency, SRL behavioral indicators, and bias-correction mechanisms to support fair teacher evaluation. Future research should employ cross-cultural qualitative designs to examine how AI-driven evaluation tools function in diverse school contexts and to ensure alignment with UNESCO’s global education equity agenda.