AIQ
AIQ
Part of the AI in Healthcare lesson guide. Teaching a different grade? 🌈 Explorer (5–7) · 💻 Hacker (11–14) · ⚡ Architect (15–18)
Open with energy but give this age band a real claim to react to — they're old enough to be a little skeptical, which is useful.
Say: "AI is saving lives right now — spotting diseases doctors miss, and discovering new medicines. Let's explore how AI is transforming healthcare!" Then ask: "Raise your hand if you think a computer could ever spot something a doctor missed. Keep it up if you think that sounds impossible." Take a quick show of hands either way — you're not grading the answer, you're building curiosity for the activity to resolve.
Bridge into the lesson: "By the end of today you'll know three real jobs AI already does in hospitals and labs — and one big rule about who's still in charge. Let's open the lesson."
If your class has already done earlier AIQ lessons, it's worth a quick callback here: remind them that every AI they've studied so far learns by studying huge numbers of past examples, not by "thinking" the way people do. That's the same idea this lesson applies to medicine — an AI that spots disease in an X-ray learned to do that by studying thousands of X-rays that doctors had already diagnosed, not by understanding illness the way a doctor does.
Builder mode works through the same three scenes as the full lesson — Diagnosis, Treatment, and Patient Care — each with four items where AI is doing the work and a plain everyday tool paired alongside it. Have students open the lesson on their own devices and walk through the scenes at their own pace, or drive it as a class on a shared screen; either way, pause after each scene to check understanding with the prompts below.
Scene 1 — Diagnosis. Students will see: an X-ray AI that can detect lung cancer with high accuracy, sometimes performing as well as a radiologist; an eye-scan AI (built by Google) that checks retina photos for diabetic blindness, useful in places without many eye doctors; and a skin-cancer AI trained on more than 130,000 images that can spot melanoma about as accurately as a dermatologist. Paired against those: a plain stethoscope, which just amplifies heartbeat sound with basic physics — no AI at all. Before revealing the fact for the skin-cancer example, try a quick predict-first question: "Do you think a computer looking at a skin photo could really spot cancer as well as a trained doctor? Thumbs up for yes, thumbs down for no." Letting students commit to a guess before the reveal tends to make the actual fact land harder than simply reading it aloud. After the scene, ask: "What do all three AI tools have in common?" Guide them toward: they all look at an image and compare it to patterns learned from many other images.
A good follow-up for this scene: ask students to guess how many X-ray images they think a computer would need to see before it gets good at spotting a pattern — take a few guesses, then reveal that real systems like these are typically trained on tens of thousands to hundreds of thousands of images (the skin-cancer example used more than 130,000). This number is usually much bigger than students guess, and it's a good concrete way to show that "learning" for a computer means seeing an enormous number of examples, not just a handful.
Scene 2 — Treatment. Students will see: an AI that helped discover a new antibiotic in 2020 by screening 100 million chemical compounds — work that would have taken humans decades by hand; a gene-therapy AI that helps design cancer treatment personalized to a patient's own tumor DNA; and surgery robots that give surgeons steadier hands and a magnified 3D view during an operation. Paired against those: a plain band-aid — just adhesive and padding. Ask: "100 million compounds is a huge number. Why would a computer be better suited to that search than a team of scientists working by hand?" Steer toward speed and scale, not "smarter" — the computer isn't more clever, it's just able to check enormously more options in the same amount of time.
It's worth being clear with this age group that finding a promising compound is only the first step. Say: "Finding a compound that looks promising on a computer is like finding a good recipe idea — it still has to be cooked, tasted, and tested by real scientists in a lab before anyone can actually use it as medicine. The computer just makes the search for that recipe idea much, much faster." This heads off the common assumption that AI "invented a cure" rather than narrowed a search.
Scene 3 — Patient Care. Students will see: smartwatches that use AI to detect irregular heartbeats and alert the wearer before they even notice symptoms; an AI chatbot (like Woebot) that offers cognitive-behavioral-therapy-style techniques for mental health, available any time of day; and AI that reads patterns across millions of health records to predict who might get sick before they do. Paired against those: a plain thermometer, which just measures temperature. Ask: "Would you trust an app to be your only mental health support, or would you want a real person involved too?" This previews the teamwork idea without stating it outright. Expect a mix of answers here — some students will say the app is fine for a quick check-in, others will say they'd always want a real person for something serious. Both are reasonable, and the useful teaching moment is asking students to say why, since that "why" usually lands on trust, judgment, or human connection — things this lesson's quiz is ultimately pointing at.
As you move through the three scenes, keep drawing attention to the "isAI: false" items — the stethoscope, band-aid, and thermometer. Students at this age sometimes assume that anything used in a modern hospital must involve AI, and the paired plain-tool examples are there specifically to break that assumption. A useful running question after each scene: "What made you sure that one didn't use AI?" Good answers point to the tool doing one fixed, simple physical job with no learning or pattern-recognition involved — physics and materials, not data.
Encourage students to notice a pattern across all three scenes: every single AI example involves comparing something new (an image, a chemical compound, a heartbeat) against a huge number of past examples it already learned from. Naming that pattern themselves, rather than being told it, is the real goal of this activity — the specific facts (94% accuracy, 100 million compounds) are memorable hooks, but the pattern underneath them is the actual takeaway.
These work well as a think-pair-share: give students 30 seconds to think, a minute to discuss with a partner, then invite a few pairs to share with the whole class. That structure tends to get more thoughtful answers from this age group than cold-calling.
This age group's quiz uses slightly harder wrong-answer options than Explorer's — rather than obviously silly choices, the distractors are things that sound plausible but overstate what AI actually does ("replace all doctors," "cure any disease"). When walking through wrong answers, it's worth naming why they're tempting, not just that they're wrong: they're the kind of overstated claim a flashy headline might make.
Close with: "Today you learned that AI can spot patterns in X-rays, search through millions of chemical compounds, and keep an eye on your heartbeat — but in every single case, a human doctor is still part of the decision. AI is a teammate, not a replacement."
Extension activity (15–20 minutes): In pairs, have students design a one-page poster for an imaginary "AI Health Helper" of their own invention. It must include: what job it does (pick one: diagnosis, treatment, or patient monitoring), what kind of data it would need to learn from, and — non-negotiable — one sentence explaining what the human doctor still does that their invention doesn't. Have a few pairs present, and use the human-doctor sentence as a quick comprehension check for the lesson's main idea.
If you'd like a shorter alternative, run a quick class debate instead: split the room into two sides and have one side argue "AI should be allowed to make the final call on a diagnosis without a doctor checking it" while the other argues against it. Give each side two minutes to prepare using examples from the lesson, then two minutes each to present. This works well as a 5-minute closer if the poster activity doesn't fit your remaining class time, and it puts the lesson's central "teamwork, not replacement" idea directly into students' own words.
If a debate side leans too hard toward "AI should decide alone," it's fine to step in with a concrete counter-scenario rather than simply declaring the answer: "Imagine the AI's training photos happened to include very few examples of a rare skin condition. What would it likely do when it finally saw one?" Most students will correctly guess it would probably miss it or misclassify it — a natural, hands-on way to surface why oversight still matters, without needing the word "bias" yet.
Timing note: the in-app lesson content plus the scene debriefs above typically runs 12–15 minutes for this age group; the poster activity adds another 15–20 minutes on top of that, so plan for roughly a 30-minute block if you're running the full extension. If you're short on time, the debate alternative compresses the same core idea into about 10 minutes total.