What My Worst-Behaved Class Taught Me About AI Rules

As a young teacher in northern Zambia, I was always handed the worst-behaved classes. Every term, the same setup: thirty teenagers, zero interest in my rules, and a head teacher watching to see if I'd survive. By the end of every term, those classes were the best in the school.

Here's what I learned: rules imposed from above get tested. Rules built with the room get followed.

My first month, I posted a long list of prohibitions. The students treated it as a challenge menu. So I changed tactics. We sat down and I asked: what kind of classroom do we want to be? They named it — focused, respectful, no one left behind. Then we built the rules together, as the price of the room they said they wanted. Compliance followed because the standard was theirs.

Schools are writing AI policies the old way right now: long lists of prohibitions handed down from the office. Students will treat them as a challenge menu.

Do what worked in my classroom. Bring students into the conversation. Ask what kind of learning community they want to be, then build the AI norms as the price of that community. A policy students helped write is a policy they'll defend.

Guardrails work best when the people inside them helped build them. That's true in a classroom in Zambia, and it's true in a school rolling out AI anywhere.

I explain frameworks; I don't give legal counsel.