The Human-Centred AI Framework for Schools

The problem the framework addresses

Human-centred AI is not achieved by adding a disclaimer below an automated decision. It requires product boundaries that preserve agency, make evidence visible and protect those who cannot meaningfully consent to data use.

The framework

• Human approval before classroom action.
• Minimum necessary learner data.
• Transparent source and confidence.
• Age and context appropriate use.
• A route to challenge, correct and remove information.

How it applies inside Asmbly

Asmbly can use AI for drafting, organisation and signals while preventing it from making high-stakes learner decisions. The teacher should always see and edit the output before it enters classroom or reporting work.

The boundary

Human review can become symbolic if the interface pressures users to accept output quickly. Design must give enough time, context and control for meaningful review.