AI literacy isn't a ban or a free-for-all. It's a skill.

For two years the conversation about AI in schools was mostly about access: block it, allow it, or quietly pretend it wasn't happening. That argument is losing its energy, because students are using these tools whether or not a policy exists. The more useful question — and the one more districts are now asking — is what students actually need to understand about AI in order to use it well.

A report from the Associated Press in August described how Charleston County School District in South Carolina spent a year consulting teachers, students and parents before rolling out AI literacy instruction this school year. Notably, the district built its own framework because its state hadn't published guidance. That pattern — schools moving ahead of policy — is now common.

Start with the flaws

What's striking about the classroom examples in that report is how much of the teaching starts with AI getting things wrong. One trainer's demonstration was simply asking a chatbot to draw a map, then letting students find the misspelled countries. The lesson wasn't "AI is bad"; it was that you can't take an answer and move on. Verification is the skill.

That line deserves to be printed on a wall. A student who reaches for a chatbot before thinking has not learned AI literacy, however fluent their prompting is. A student who can say "this is a task where the tool will make me worse" has.

What a real programme covers

Our own AI Literacy programme is built on this idea, and its arc is a useful checklist for anyone designing one. Students move from using AI tools to questioning their outputs, evaluating them for bias and accuracy, creating something original with AI as a collaborator, and finally deciding — forming a position on responsible use and acting on it. The order matters. Evaluation before creation. Judgement before scale.

• Foundations: what these systems are, and how rule-based software differs from machine learning.
• Prompting as a craft — and prompt logs as evidence of thinking.
• Bias critiques: finding where an output is skewed, and explaining why.
• Ethics and citizenship: privacy, data, and the dilemmas that don't have a clean answer.
• A capstone that students design, build and present to a real audience.

None of this needs a computer science department. It runs on Chromebooks with free tools, and it works best when the person at the front of the room is a thoughtful teacher rather than a technologist. The goal is students who can think alongside a machine without handing over the thinking.