AIQ
AIQ
Part of the AI or Not? The Game lesson guide. Teaching a different grade? 🌈 Explorer (5–7) · 💻 Hacker (11–14) · ⚡ Architect (15–18)
Builder mode students (Axiom's crew) respond well to a challenge framed as a test of their instincts, not a lecture. Open with energy and a direct callout to how tricky this round is:
"Let's play a game! I'm going to show you some tricky examples. Some of them SEEM smart, but they're not actually AI. Others use AI in ways you would never guess. Can you tell the difference?"
Before opening the app, put two quick examples on the board and ask for a show of hands: "Who thinks a basic calculator app counts as AI?" and "Who thinks the thing that sorts your email into 'Primary' and 'Promotions' counts as AI?" Don't correct anyone yet — just note the split on the board. Tell them: "By the end of this lesson, you'll have one test that answers both questions, and I bet it changes at least one of your votes."
Give them the actual test up front, since Builder-age students do better working with a clear rule they can apply themselves rather than discovering it purely by trial and error:
"The test is: does it LEARN and get better over time? Something that just follows the same fixed rule every single time — no matter how digital or fast it is — is NOT AI. Something that adapts and improves the more data it sees IS AI."
Have a student repeat the test back in their own words before moving on, to make sure it's landed before the examples start coming fast.
One more warm-up move that works well at this age: ask "Can something be AI and also be really simple to use?" Most Builder-age students assume "AI" means complicated to operate, when really the opposite is often the goal — a well-built AI feature usually looks effortless from the outside precisely because the learning happens invisibly, behind the scenes, before the student ever taps anything. Flag that they'll see this pattern repeatedly in today's examples.
Open the lesson and go through the app's three rounds as a class first, using the "does it learn?" test out loud for each item, before letting students play the sorting practice round independently.
Pair the calculator app with the auto-DJ feature some music apps have. "A calculator app just follows math rules — 2 + 2 is always 4, forever, no matter how many times you ask it. It's not AI. But an auto-DJ feature, like the one in Spotify, uses AI to mix songs together and try to guess your mood from what you've been listening to. Same category — 'app on your phone' — completely different answer, because one learns from your behavior and one doesn't."
Pair the digital clock with the voice-changer filter. "A digital clock counts a crystal vibrating a fixed number of times per second — precise, but it's not learning anything, it's just counting. A voice-changer filter that transforms your voice in real time, though, is often running AI models to reshape the sound convincingly. Ask students: which one is easier to build — the clock or the voice changer? (The voice changer, by far, is the harder engineering problem, which is part of why it needs AI and the clock doesn't.)"
Pause here and ask the class to predict a rule from these two pairs before moving on: "Notice both 'not AI' examples — the calculator and the clock — are things that just do arithmetic or counting. Both AI examples are doing something with sound or music that would be really hard to write as a simple rule. Why do you think tasks like 'recognize what mood this song has' or 'reshape this voice convincingly' are harder to hand-code than tasks like 'add these two numbers'?" Guide them toward the idea that some tasks (arithmetic) have one clear correct answer and a simple rule to get there, while others (mood, sound, style) don't have one obvious rule — which is exactly why they benefit from a system that learns patterns from lots of examples instead.
This round is the heart of the lesson for this age group — it's where "looks boring, is secretly AI" lives. Walk through Gmail's spam and inbox sorting: "Gmail uses AI to figure out which of your emails are spam and which are important, and it gets better at this the more email it processes — both yours and everyone else's. Compare that to an LED bulb, which just converts electricity into light. Efficient, sure, but there's no learning happening — it's not becoming a 'better' bulb over time."
Introduce the phone battery example, which tends to surprise this age group the most: "Modern phones use AI to learn your usage patterns — which apps you open at 7am, which ones you never touch — and manage battery power accordingly. It's genuinely learning from your specific habits, which is exactly the test we're using." Contrast with a calculator watch: "Same idea as the calculator app — it's just a tiny calculator on your wrist. Math, not machine learning."
Use this round to stretch their sense of where AI shows up. Smart farming: "Farmers now use AI-equipped drones to fly over fields and check crop health from the air — spotting diseased or stressed plants long before a person walking the rows would notice." Smart toilets: "Yes, this is real — some toilets on the market use AI to analyze health data. It sounds like a joke, but it's a genuine, if unusual, product category." Contrast both with a stapler: "It hasn't functionally changed in over a hundred years. Push down, staple goes through paper. No learning, no AI, ever."
Finish with AI-generated art: "AI can create paintings, music, and stories from nothing but a written description. That's AI operating in a completely different domain from spam filters or battery management — same underlying idea (learning from huge amounts of data), very different output."
Take a moment here to have students connect all four "surprise" examples back to the core test: farming drones, health-sensing toilets, battery managers, and AI art generators look nothing alike on the surface, yet all four pass the exact same test ("does it learn from data and improve?"), while a stapler, a calculator, and an LED bulb — also nothing alike on the surface — all fail it for the exact same reason. This is the moment to name explicitly that the test is what matters, not any surface-level similarity between examples.
Once the guided pass is done, let students work through the app's sorting practice round on their own (or with an opt-in squad partner) to classify the remaining items — auto-DJ, email sort, battery AI, smart farm, calculator, calc watch, stapler, LED bulb — without your narration.
Close by returning to the board vote from the hook. Ask again: "Calculator app — AI or not? Email sorting — AI or not?" and see how the class's answers have shifted, then restate the test one more time: "Does it learn and get better over time? If yes, AI. If no, it's just rules — no matter how fast, fancy, or digital it looks."
Extension activity — Squad AI Audit (15–20 minutes): Split the class into small squads (Builder mode's opt-in squad feature makes this a natural fit if your students already use it in-app). Give each squad five minutes to list every app, device, or gadget they can think of from home or school, then sort each one into "AI" or "Not AI" using the test — and for every "AI" one, they must also write down a guess at what data it's learning from (e.g., "the music app learns from which songs I skip and which I replay"). Bring the squads back together and have each one present their most surprising find in both directions. Award bonus XP or a badge shout-out to any squad that finds a genuinely surprising real example the app itself didn't cover.
If your class has extra time, add a second phase: have each squad pick their single best "not AI, but marketed as smart" find and design a short, honest one-sentence label for it — the kind a truthful product box might carry instead of the marketing version. For example, a "smart" toaster with a timer dial might get relabeled "adjustable timer, no learning involved." This turns the audit into a small exercise in seeing through marketing language, which is a skill that will keep paying off well beyond this one lesson.