Your students are already using AI. This is a free, hands-on way to teach them how it actually works, and how to use it honestly — with nothing to install, no accounts, and no data leaving the room.
Free · No login · No accounts · No live AI · Nothing collected (COPPA/FERPA-friendly) · Standards-aligned
📄 Print the one-page Teacher Starter Kit →
Two age tracks, same honest voice — the hands-on kids rooms (ages 8–12, projectable in Presentation Mode) and the teen track (13+ — the feed, deepfakes, AI companions, and homework, honestly), which is already a station-by-station walk built to project. Both are jump-to-any-room, no login, nothing collected.
The kids lab (LearnAI4Kids) is content-complete, standards-aligned, and safe to use with students right now. The "we welcome educator feedback" notes on the site are about continuous improvement, not a safety hold. The practical details:
Nothing to install or sign into. Works in any browser — Chromebooks, tablets, phones, the projector. Just open the link.
No accounts, no logins, no tracking. The only thing stored is which rooms a student finished, kept locally on their own device. Nothing they type leaves the browser. COPPA/FERPA-friendly by design.
There's no chatbot and no AI running on the site. Every AI response a student sees is a real example, captured ahead of time and played back by the page. The activities are plain in-browser lessons (just JavaScript) that teach how AI works, so nothing a student types is ever sent to an AI.
Each exhibit ~3–5 min; each room ~15–25 min; the full lab ~2 hours, or one room at a time as a warm-up.
Assign rooms on devices and circulate, or, to walk a room on the board, use Presentation Mode: it steps through one concept at a time, big and high-contrast for the back of the room, with the narration ready to play. Advance with a clicker or arrow keys; the activities still run live on the board. (Or tap 🖥 Present on any room.)
The kids rooms are built for ages 8–12. Older students have their own teen track (13+) — the same fundamentals, pitched for high-schoolers. Ready for more depth? The adult journey.
Most teachers are. Do the 2-minute primer or the adult journey first — it makes leading the discussion easy.
Every exhibit cites its AI4K12 Big Idea on the page. Here's the room-by-room map you can paste into a lesson plan or hand to an administrator. Big Ideas: 1 Perception · 2 Representation & Reasoning · 3 Learning · 4 Natural Interaction · 5 Societal Impact.
| Room | Teaches | Big Ideas | A discussion question |
|---|---|---|---|
| 1 · What is AI? | AI predicts; it doesn't "know" | 3, 5 | What changed when you picked a less-likely word? When is a fast guess good enough? |
| 2 · The whole AI family | AI is many tools, not one chatbot | 1, 5 | Which kind of AI surprised you? Where do you meet AI without noticing? |
| 3 · When AI is wrong | Confidently wrong; why it happens | 2, 3, 4 | How could you tell it was wrong? What would you check before trusting it? |
| 4 · Is AI biased? | Bias comes from lopsided data | 3, 5 | What does "bias" mean? Where did it come from, and how would you make the examples more even? |
| 5 · Using AI well | Prompting; tutor vs. answer-machine | 4, 5 | How did changing your question change the answer? When should AI be a tutor? |
| 6 · Real worries | Deepfakes, privacy, cheating, "AI friends" | 1, 4, 5 | How can you spot a fake photo? Why isn't an AI "friend" a real one? |
| 7 · Your turn | Consolidation; the verify habit | 5 | Which super-question will you use? What will you try with AI this week? |
For older students (13+), the teen track teaches the same AI fundamentals in a voice built for teenagers — the same honesty, tuned to high-school topics (the feed, deepfakes, AI companions, privacy). The 10 teaching stations (plus an intro and a wrap-up) map to AI4K12 the same way the rooms do. Earning the Scouting America AI merit badge? See which requirements this helps with — and which it doesn't.
| Station | Teaches | Big Ideas | A discussion question |
|---|---|---|---|
| 01 · What it actually is | AI predicts the next word | 3 | If it's "just guessing the next word," why does it sound so smart? |
| 02 · Under the hood | Tokens (not letters) + how it's trained | 2, 3 | Why can't it count the R's in "strawberry"? What did the rating step teach it to value? |
| 03 · Confidently wrong | Hallucination; confidence ≠ correct | 3 | How would you check a confident-sounding claim before you trust it? |
| 04 · The feed | Recommender AI optimizes for watch-time | 5 | What is your feed optimizing for, and how could you tell? |
| 05 · Is it biased? | Bias from a lopsided training pile | 3, 5 | Where does AI bias come from, and who has to fix it? |
| 06 · Use it well | Prompting — who / what / how | 4 | How did adding context change the answer? When should AI be a tutor, not an answer-machine? |
| 07 · Homework, honestly | Learning vs. offloading; policy first | 5 | Where's the line between using AI to prep and cheating future-you? |
| 08 · Real or fake | Deepfakes (photo/voice/video) + your likeness | 1, 5 | How can your own face or voice be misused, and what's the move if it happens? |
| 09 · The friend that isn't | Sycophancy + the Eliza effect | 4, 5 | Why does an AI "friend" agree with you, and why is that risky? |
| 10 · Free vs paid & privacy | Data, training-on-you, redaction | 5 | What should never go into a chatbot, and why does "free" change the answer? |
AI4K12 is the framework we map to exhibit-by-exhibit (table above). The approach — concept-first, honest about limits, safety- and citizenship-minded — is also designed to be consistent with the other widely-used K-12 AI guidance an administrator may want to see. Every link goes to the source.
| Framework | Who's behind it | How LearnAI4Kids relates |
|---|---|---|
| AI4K12 Five Big Ideas in AI | AAAI + CSTA | Our formal spine — every exhibit cites its Big Idea (table above). |
| TeachAI AI Guidance for Schools | Code.org, ISTE, Khan Academy, ETS | Matches its "teach about AI, use it responsibly, keep a human in the loop" stance — our AI-detector guidance follows it. |
| Common Sense Education AI literacy + family toolkit | Common Sense Media | Shares its digital-citizenship and family-conversation approach; our For Parents page plays the same role. |
| UNESCO AI competency frameworks (2024) | UNESCO | Reflects its human-centred competencies for students — understand, use critically, and question AI. |
| Digital Promise AI Literacy framework (2024) | Digital Promise | Built on the same three modes we practice: understand, evaluate, use. |
| ISTE Standards + AI guidance | ISTE / ASCD | Supports the same student-as-critical-user goals in the ISTE Standards for Students. |
We map formally to AI4K12; the others are the recognized landscape this resource is built to be consistent with — listed so you can check it against whatever your school or district already uses.
A "green / ask-first / not-OK" sort for real homework situations. Teaches the line between using AI and cheating better than a blanket rule, and starts the class conversation for you.
A printable card: "Did you make this up? · What if I'm wrong? · How could I check?" Laminate it and put it on the wall. It's a verify habit students can use on any AI, forever.
The one norm that handles most AI-use questions. Pair it with the copy-paste tutor prompt (on the parents page) that turns AI from answer-machine into patient tutor.
Don't rely on AI-writing detectors. They are not reliable enough to accuse a student. They falsely flag human writing — disproportionately for multilingual students and neurodivergent writers, and they miss plenty of real AI use. Several universities and districts have turned them off for exactly this reason.
What works better than detection:
Nothing here asks you to take our word for it. If you want to check the approach against the evidence, hand a skeptical administrator the citations, or brush up on a term before you teach it. It's all in one place, and all free.
Each factual claim on the site links to its primary source — public-opinion data, energy & water figures, deepfake-fraud reports, the cognition/learning-science studies, and the "how AI actually works" papers. Primary research and government reports only, no marketing listicles.
The honest read on the learning science — London cab drivers, the Google Effect, Lee et al. (Microsoft/CMU), MIT Media Lab, and why it depends on how AI gets used. The evidence you'd want to justify the "did you try first?" norm to an administrator or a parent.
Every term the lab uses — tokens, hallucination, training, bias, sycophancy, the Eliza Effect — defined in one or two sentences, with where it shows up. Good for your own prep or to project as a word wall.
LearnAI4Kids was built by one parent — a dad of four and a non-profit IT director — who uses AI every day and wanted honest, agenda-free AI information for his own kids, but couldn't find it. No ads, no sign-up, nothing collected, nothing for sale. The full story is on About this site.