Before the policy: five questions boards should answer about AI

Artificial intelligence arrived in schools faster than governance did. The typical response has been to adopt a policy borrowed from another district, publish it, and hope that clarity follows. It rarely does, because the policy answers questions nobody in the organization actually asked.

Boards get further by settling five questions first. One: what are we protecting? Student data, assessment integrity, staff time, and instructional judgment are different risks with different controls. Naming them prevents a single blanket prohibition from standing in for real risk management.

Two: what are we trying to enable? If the answer is only efficiency, the initiative will be perceived as cost-cutting. If the answer includes better feedback for students, stronger differentiation, and reduced administrative load for teachers, the work has a constituency.

Three: who decides which tools enter classrooms? A single approval path — with security review, instructional review, and a named owner — prevents the shadow adoption that follows every prohibition.

Four: what does acceptable use look like by role? Guidance written for teachers, students, and administrators separately is followed. Guidance written for 'users' is not.

Five: how will we know it is working? Choose two or three indicators before launch — teacher time reclaimed, feedback turnaround, help-desk volume — and review them at a fixed interval.

Districts that answer these five questions typically write a shorter policy in less time, and spend the saved effort on the part that actually changes practice: professional learning, exemplars, and a clear escalation path when something goes wrong.