Strategic leadership for agentic AI in education: a conceptual framework with reference to the United Arab Emirates

Abstract

Agentic Artificial Intelligence (AI) shifts educational leadership from whether to adopt a tool to which actions a configured system may take, under whose authority, and with what capacity for intervention and remedy. Through critical integrative synthesis of educational leadership, organisational decision-making, AI governance, and agentic-AI scholarship, this Conceptual Analysis develops educational agentic alignment: a relational judgement about whether delegated machine authority fits educational purposes, human decision rights, professional capabilities, safeguards, pedagogy, and public-value obligations. The focal unit is a dated and versioned educational delegation arrangement. The framework separates agentic capability, use-related exposure, and usable governance capacity; organises leadership decisions through seven domains across four levels; and proposes three mechanisms concerning safeguard shortfall, responsibility–capacity mismatch, and evidence-linked realignment. Each proposition specifies the analytical unit, variables, expected direction, measurable outcomes, time order, moderators, rivals, and evidence that would count against it. Official evidence from the United Arab Emirates illustrates contextual boundaries without implying national implementation or framework validation. The contribution is an education-specific explanation of fit and misfit in human–machine authority, together with a practical basis for bounded decisions and comparative research.