Is AI Fair? When the Examples Are Skewed
1. What your child learns and why
Your child “teaches” a simple machine to tell apples from pears. When it is shown 10 apples and only 2 pears, the machine calls every new pear an apple. Your child adds examples and sees when the mistake disappears. This is how they understand bias in artificial intelligence: the machine is not bad, it copies its examples — and the gaps in them. At the end they discuss with you who should check when a machine suggests decisions about people.
2. Three questions for dinner
- “Why did the robot get the pears wrong?”
- “Where else might a machine see examples of only some people?”
- “Who should have the last word when something is decided about a person?”
3. 10 minutes together without a screen — the “bird machine”
- Draw 6 slips with pigeons and sparrows (“bird”) and 2 with a cat and a dog (“not a bird”).
- One person is “the machine”: it learns only from these and states its rule. Then show it a penguin, an ostrich and a duck.
- Add these examples, “teach” the machine again and compare the rules.
4. What success looks like
- Your child explains that the machine is wrong about the kind it saw few examples of.
- They say the fix is more and varied examples, and for decisions about people a person checks.
5. Common difficulties
- “So AI is bad.” → “It isn't bad — it learns from what we give it. Who is responsible for the examples?”
- The topic of people feels heavy → go back to the fruit and the birds; there is no need to discuss real cases.
Say: “You fixed the machine — that is what engineers do.” Praise the checking, not the quick answer.
6. Safety
All examples are fruit and birds — no people, faces or photos. Everything on screen is a game in the page; nothing is typed or uploaded. No chatbots and no registration. The talk about decisions about people is general and led by you, without blaming any group. The legal points follow an unofficial copy of the AI Act ⚠️ (not checked against EUR-Lex); this is not legal advice. If something scares you — tell an adult you trust.
7. Where to go next
Previous lesson: DB-II-02 “Why Is It Showing Me This? Recommendations”. Next: DB-II-04 “A Picture from Words: How AI Draws and Whose It Is”. See also DB-IB-06 “How AI “Guesses” — and When It Is Wrong”.
You don't need to know the answer. Say: “Let's find out together.”
VERIFIED · 10.11.2026 · AI help: yes · reviewer: awaiting teacher and psychologist · East Accelerator Foundation · free · licence CC BY-SA 4.0