AI LITERACY FOR AGES 12–16

Check,
then trust.

Unpack trains teenagers to pause before they believe an AI — and then go find out. Hands-on missions, run by your teachers, private by design.

Request a pilot See how a mission works
Not anti-AI. Pro-thinking.
try it · case 03 / historyAI in use
you ›how long did it take to build the eiffel tower?
ai ›The Eiffel Tower took about 7 years to build. Construction finished in 1895, just in time for the World’s Fair in Paris.
■ PAUSE
Sounded sure. Do you believe it?
THE PROBLEM

The enemy isn’t AI. It’s passive trust.

Teenagers already use AI for homework and for looking things up. Most answers arrive fast, fluent and sure of themselves — and most people never check them.

~64%of US teens use AI chatbots — top uses are searching for information and schoolwork.¹
Friendliersounding AI gets trusted more — even when it’s wrong.²

¹ Pew Research Center, 2025–2026. ² Penn State research on AI tone and trust. Heavy reliance is linked to “cognitive offloading” and what MIT researchers call “cognitive debt.”

HOW A MISSION WORKS

Every AI answer is a compressed version of the truth. Students learn to unpack it.

  1. 01BriefA voice guide sets the case.
  2. 02AskThey question a real AI in a terminal.
  3. 03The PauseProgress stops. “Do you believe this?”
  4. 04VerifyThey search real sources and compare.
  5. 05CatchThey log what’s wrong — with proof.
  6. 06ReflectTheir gut call vs. what they found.

The moment that matters is the Pause. We don’t tell students the AI is wrong — we make them decide, then go find out. Sometimes it lied. Sometimes it was right. Either way, they know why.

SEVEN KINDS OF MISSION

Seven ways an answer goes wrong. One where it doesn’t.

The Confident FibPlausible, fluent, wrong. Find the one fact that doesn’t hold.
The Tidy Wrong AnswerClean reasoning, wrong result. Spot the step that breaks.
Your Voice FirstWrite your own ideas before the AI is allowed in. Then compare.
Headline vs. StoryA news summary that drops a caveat — or invents a detail.
The Leading QuestionSame question, two framings, two answers. Who’s steering?
The Source HuntTrace a claim back to where it really came from — if anywhere.
AND THE TRAP“Actually Right.”Sometimes the AI is correct — and students have to prove that too. We’re training calibration, not cynicism. Doubting everything is just another way of not thinking.
FOR TEACHERS

Run a two-hour session without writing a single question.

01 · COMPOSEPick the mixChoose mission types, themes and age band. Every returning student gets fresh cases — no repeats, no copying answers.
02 · RUN LIVESee the roomWho’s stuck at the Pause, who hasn’t opened a source, who’s ready to race a partner. Nudge without giving the answer away.
03 · REVIEWKnow who needs whatWho caught what. Who trusted too easily. Who doubted without evidence. Next session adjusts itself.
One session, two hours:
Hook · 15 min
A confident AI, live, being wrong
Missions · 70 min
Students work through cases at their own pace
Reflection · 20 min
Alone vs. with AI
Teach it back · 15 min
Explain one thing — no AI
WHAT STUDENTS LEAVE WITH

AI as an assistant. Never the authority.

  1. 01Spot the pause — notice the moment they’re about to accept an answer.
  2. 02Verify independently — check claims against real, primary sources.
  3. 03Detect errors and limits — fake facts, fake citations, flawed reasoning, framing.
  4. 04Protect their own voice — ideas first, AI second.
  5. 05Research like an investigator — trace claims to where they started.
  6. 06Understand to own it — explain a verified idea without AI.
PRIVATE BY DESIGN

Nothing a student types is ever used to train an AI.

That’s built into the architecture, not written into a policy someone can change later.

Pseudonyms only. Students are “fox-27”, never their real name. No location. No unnecessary fields. Ever. Consent first. Parental or school consent before any student joins. Strictest by default. Schools can tighten settings — never loosen them. Honest about AI. Students are told plainly when they’re talking to one. Deletion that means it. Requests are handled in the dashboard and logged.
Read how privacy works

Questions we get asked

Is this teaching kids that AI is bad?

No. The message is “check, then trust.” Some missions have an AI that’s completely right — students who doubt it without evidence learn as much as students who trust a wrong answer.

What age is it for?

12 to 16, in two bands. 12–13 gets simpler claims and gentler stakes; 14–16 gets subtler traps.

What do we need to run a session?

Two hours, one device per student with a browser, and headphones for the voice guide (captions are always on).

Do students need accounts or email addresses?

No. They join with a class code and get a pseudonym. Your school keeps the list of who’s who.

Who writes the missions — and are they accurate?

Missions are drafted, checked against real sources, and approved by a person before they go live. Time-sensitive news missions expire automatically. The bank is refreshed every six months.

How much does it cost?

It depends on your school and how many classes take part. Get in touch and we’ll talk it through.

Start with one class.

Run a pilot session. See where your students pause — and where they don’t.

Request a pilot