Breaking urban-rural educational inequality: application of an AI relative fairness model in teacher performance evaluation and bonus allocation

Abstract

Teacher performance evaluation and bonus allocation in Chinese schools still rely heavily on subjective judgment, and the resulting distortions disproportionately affect rural teachers, exacerbating compensation disparities and attrition that undermine the rural revitalisation agenda and SDG 4. This study asks how artificial intelligence (AI) could make these processes fairer. Following the PRISMA 2020 guidelines, a systematic search of five databases for 2010--2025 returned 4,713 records, of which 30 studies were included and synthesised using structured thematic coding and keyword co-occurrence mapping. Three overlapping themes emerged: AI applications relevant to evaluation (coded in 60% of the included studies), fairness pathways such as explainable AI and participatory decision-making (80%), and urban-rural inequality (20%). The reviewed literature indicates that AI-assisted evaluation can improve the consistency and evidential basis of teacher assessment and relieve administrative burden, but that these gains are conditional on confronting algorithmic bias, model opacity, privacy risk, and the rural digital divide; evaluation models designed specifically for teachers, let alone adapted to rural Chinese contexts, remain scarce. On this basis, the paper proposes a four-dimensional relative fairness model (4DG) combining data-integrity safeguards, transparency through explainable and generative AI, teacher participation, and ethical governance with auditable records. A deliberately simple, assumption-driven simulation - reported transparently as such - illustrates that correcting a substantial share of evaluation bias could narrow the urban-rural bonus gap by roughly 15-25%. Because the review's quantitative synthesis could not be verified against reproducible study-level records, no pooled effect estimate is presented as an empirical finding; the paper's contribution is the governance framework and the agenda it sets for empirical piloting in China's urban-rural context.