Co-Designing Educational AI: An Evidence-Based Handbook for Teachers, Students, and Institutions—from Purpose to Stewardship

Educational AI should be built with the people whose learning, work, data, rights, and institutions it changes. AI applications are entering classrooms, campuses, assessment systems, advising, learning analytics, and institutional decision-making. Yet many are still designed around technical possibility rather than educational purpose-and presented to teachers and students only after the consequential choices have already been made. Co-Designing Educational AI is a scholarly and practical handbook for educators, students, school and university leaders, administrators, researchers, designers, developers, and policy teams. It synthesises evidence from participatory design, AI in education, learning sciences, responsible AI, implementation research, and technology governance. At its centre is PARTICIPATE-AI, an eleven-stage lifecycle for deciding whether AI is warranted, sharing decision rights, translating stakeholder knowledge into system requirements, allocating human-AI roles, evaluating learning and distributional effects, and governing applications after release. Readers will learn how to: - distinguish consultation from meaningful stakeholder influence;- turn educational values into testable design requirements;- co-design data, models, interfaces, workflows, and governance;- evaluate learning, equity, workload, accessibility, safety, and trust;- create evidence and decision trails that remain auditable; and- monitor, contest, adapt, suspend, or retire deployed systems. The result is an evidence-based blueprint for building educational AI with people-not merely for them.