AI-Assisted Decision Support Systems for Educational Policy Planning

Abstract

Educational policy planning -- decisions about curriculum design, resource allocation, teacher training investment, school system restructuring, and technology adoption -- is consequential, complex, and data-intensive, yet typically made by human policymakers with access to limited, delayed, and poorly integrated data. AI-assisted decision support systems (AI-DSS) for educational policy provide policymakers with evidence synthesis, scenario modelling, causal inference from large educational datasets, and impact prediction that augment human judgment without replacing it. This paper proposes the AI Decision Support for Education Policy Framework (AIDSE-PF), a systematic design and evaluation of five AI-DSS architectures -- evidence synthesis engines, causal impact simulators, resource optimisation AI, equity impact assessors, and natural language policy assistants -- applied to six education policy domains: dropout reduction, curriculum reform, teacher allocation, technology adoption, equity gap closure, and learning outcome prediction. AIDSE-PF is evaluated through deployment at three national/regional education ministries over 18 months and validated against subsequent policy decision quality and learner outcome improvements. Key results: evidence synthesis engines reduce policy information processing time by 68.4%; causal impact simulators improve policy impact prediction accuracy by 42.4% vs. baseline expert judgment; equity impact assessors identify demographic disparities missed in 72.4% of standard policy analyses. The framework introduces the Policy Decision Quality Score (PDQS) and provides an AI-DSS deployment roadmap for educational ministries.