<![CDATA[Artificial intelligence (AI) is increasingly used to automate administrative work, support institutional decision-making, optimise resources, and improve communication in educational organisations. Yet the empirical evidence remains fragmented, dominated by perception studies, and methodologically heterogeneous. This study synthesised empirical research published from 2020 to 2025 on AI-based educational administration and conducted an exploratory meta-analysis of statistically compatible outcomes. A structured search of scholarly indexes, publisher repositories, and citation networks identified 11 empirical studies covering school and higher-education settings. The narrative corpus represented more than 2,400 respondent- or school-level units and 150 administrative records. Only two studies (combined N = 260) reported sufficiently compatible statistics for standardised effect-size estimation. Using the author-reported pre–post effect, a random-effects model yielded Hedges’ g = 1.00, 95% CI [0.24, 1.77], with substantial heterogeneity (I² = 90.2%). A sensitivity analysis reconstructing the pre–post effect from the reported t statistic produced a more conservative pooled estimate of g = 0.65, 95% CI [0.45, 0.86]. Narrative findings indicated potential improvements in processing time, reporting accuracy, workflow coordination, decision support, and resource allocation. However, most studies were cross-sectional, single-site, perception-based, or lacked control groups, and one effect estimate showed internal statistical inconsistency. The evidence therefore supports a cautiously positive conclusion: AI can enhance educational administration when it is embedded in redesigned workflows, staff development, human validation, and risk-based governance, but the current evidence base is insufficient for strong causal or universal claims. Future research should use controlled multisite designs, standardised administrative outcome measures, longitudinal evaluation, transparent system descriptions, cost-effectiveness analysis, and explicit auditing of privacy, fairness, explainability, and accountability]]>
