Schools need an AI policy. That part is obvious now.
What's less obvious is what comes after the policy. A document tells students what they can't do. It doesn't teach them what they should do. And for most schools, the honest answer is that we're still figuring that part out ourselves.
This year, my school started using Flint K-12. Building this course was my attempt to answer the question: what would it actually look like to teach students how to use AI well?
What Flint is
Flint is an AI platform built specifically for K-12 schools. The core idea is different from a general chatbot. Instead of handing students direct answers, the AI tutor called Sparky asks guiding questions, gives hints, and walks them through their own mistakes. It's Socratic by design. The point is to build understanding, not to shortcut it.
A few things that matter for schools:
- Students get a personalised tutor that adapts to their pace and level, available whenever they're working.
- Multilingual support across a wide range of languages and dialects, useful for language practice and for students who are still building English proficiency.
- Role-play scenarios debating a historical figure, practising a conversation in a target language, presenting an idea to a character from literature.
- A walled-garden environment that complies with COPPA and GDPR, with student data protected and school conversations not used to train public AI models.
- AI literacy built into the design, so students learn how to prompt, analyse, and work with AI rather than bypass it.
The last point is the one that changed how I thought about this. The platform isn't just a tool students use. It's a space where they can learn what AI is, what it's good at, and where it goes wrong.
Why I took the AI for Education course first
Before building anything for students, I took the Essential Guide to AI from AI for Education. It's aimed at teachers, and it gave me the foundation I was missing, what generative AI actually does, where the risks are, and how to think about it in a classroom context.
I'd recommend it to anyone in a school leadership role who feels like they're being asked to make decisions about AI without enough background to make them well. It's the kind of course that makes you realise how much of the conversation happening around you is incomplete.
The course I built
After finishing that, I built a student-facing AI literacy course on Flint. It's now in the public library, which means any teacher on the platform can use it.
The course runs across six lessons:
Lesson 0 AI Choices: Think, Check, Create
An introduction to the three habits that run through the whole course.
Lesson 1 Your Journey to AI Literacy Begins
What AI is, what it isn't, and why literacy matters more than tools.
Lesson 2 How Generative AI Works
A plain-language explanation of what's happening under the surface.
Lesson 3 Level Up Your GenAI Prompting
Practical skills for getting better results, and understanding why prompts work the way they do.
Lesson 4 Spotting GenAI Mistakes and Staying Safe
Hallucinations, bias, misinformation, and the guardrails students need.
Lesson 5 Your Personal GenAI Code
Students write their own set of principles for how they'll use AI.
Each lesson is designed to run in about fifteen minutes. That's short enough to fit into a tutor period or a weekly slot without disrupting anything else.
How we're running it at school
We spend fifteen minutes a week on the next lesson. It's not a big time commitment, and it doesn't require a specialist teacher, any form tutor can run it.
When students finish the course, they design their own Personal AI Code on Canva. It's a one-page document: their own rules for how they'll use AI responsibly, in their own words, with their own design.
Then we print them.
They go up on the classroom wall. Thirty students' worth of AI principles, in their handwriting and their visual style, visible every day. It's a reminder, but it's also a statement, this is what we agreed to, together.
What I learned building it
Putting this course together taught me more about AI than any article I'd read. Writing for students forces you to explain things clearly. You can't hide behind jargon when your audience is fourteen years old.
It also made me realise how much of the current conversation is backwards. We spend a lot of energy on detection and prohibition. Less on teaching students what good AI use looks like. But students are going to use these tools regardless. The question is whether they'll do it thoughtfully or carelessly.
A policy sets the boundary. A course teaches the skill.
You need both.
If you're building something similar at your school and want to compare notes, or if you've taken the AI for Education course and want to talk through how you've applied it get in touch. I'd genuinely like to hear what's working elsewhere.