› Grades 3–5 · ages 8–11 · Structured debate
Who taught the robot? Is it fair?
“A robot that learned from unfair examples can still make fair decisions.”
Students discover that AI learns from examples, then argue about whether a machine trained on lopsided examples can ever be fair. A hands-on card sort makes bias concrete before the first mini-debate of the year.
50 minutes · one class period · free to photocopy
Students will be able to
- Explain that AI learns patterns from the examples it was shown.
- Predict what a robot will get wrong if its examples were missing a group.
- Argue a side in a short, structured mini-debate with one reason and one example.
Vocabulary
- Training data — The examples a model learned from — often the whole internet, licensed corpora, or both.
- Bias (in data and models) — Systematic skew — who’s represented in training data and who gets hurt in outputs.
- Artificial intelligence (AI) — Software that performs tasks we used to assume required human judgment.
- Human in the loop — A person reviews or approves before action goes live.
Agenda · 50 minutes
| Minutes | Segment | What happens |
|---|---|---|
| 0–8 | Warm-up | All-dogs deck. Establish: the robot learns from what it's shown. |
| 8–16 | Vocabulary | Read the training-data and bias cards. Class rewrites each in ten words or fewer. |
| 16–22 | Team prep | Split into Team Fix-the-Data and Team Keep-a-Person. Each team lists three reasons on the organizer. |
| 22–40 | Mini-debate | Opening (1 min each), huddle (2 min), rebuttal (1 min each), open questions with hands (6 min), closing (30 sec each). Audience votes on best reasons. |
| 40–50 | Debrief + exit ticket | 'When would you want a person checking?' Fill in the handout. |
Fix the data — the robot can be fair
If you show the robot fair examples, it will learn fair patterns.
- The robot isn't mean; it just needs to see everyone.
- Adding the cat card fixes the pet problem.
- A robot doesn't get tired or play favorites like people do.
Keep a person — the robot can't be trusted alone
You can never show it every example, so a person must check the important stuff.
- There's always a cat you forgot.
- The robot can't tell you when it's unsure.
- If it decides something big — like who gets picked — a person should look too.
Discussion questions
- 1.openingWas the all-dogs robot being unfair on purpose?
- 2.coreHow many cat cards would it take before the robot 'knows' cats?
- 3.coreWho decides which examples the robot sees? Does that matter?
- 4.coreName a decision you'd let a robot make alone. Name one you wouldn't.
- 5.closingWhat is the fairest way to teach a robot?
How structured debate runs
Two teams, assigned sides. Opening statements (2 min each), rebuttal huddle (3 min), rebuttals (2 min each), open cross-fire with hands (5 min), closing statements (1 min each). The audience votes on which side argued better — not which side they agree with.
In grades 3–5
Short rounds, concrete examples, lots of moving. Fairness and friendship are the live wires.
Preview the full pack
Cover, teacher guide, and every student handout. Print teacher pages once and student pages one per learner.
On a phone, open the PDF in a new tab — in-page previews are unreliable in mobile Safari.
Standards
- CCSS ELA SL.3.1 / SL.4.1 / SL.5.1
- CCSS ELA W.4.1 / W.5.1 (opinion with reasons)
- CSTA 1B-IC-18 (impacts of computing)
More for grades 3–5
Taught it? Tell us how it went: send classroom notes · AI concepts field guide · Try the Activities Pass free · Commission a full unit