What is AI bias? (Explained for kids and parents)

AI bias is when an AI works better for some people than others — usually because it learned from training data that didn't represent everyone fairly. AI bias isn't intentional, but it's a real problem and an important thing for kids to understand.

How to explain it to a 7-year-old

🧒 "If you teach a computer using only pictures of cats with stripes, it'll think cats must have stripes. When it sees a calico cat, it'll be confused. That's bias — the AI was trained on a wrong-shaped slice of the world."

How to explain it to a 14-year-old

🎒 "AI bias happens when training data over-represents some groups and under-represents others. The model performs well on the majority case and poorly on the rest. Famous examples: face recognition working worse on darker skin, hiring AI rejecting qualified women, voice assistants struggling with non-American accents."

Three real-world examples

1. 🟫 Face recognition — early systems worked at ~99% accuracy on white faces and ~65% on Black faces because the training data was overwhelmingly white. Joy Buolamwini's research forced major companies to fix this.
2. 🎯 Hiring AI — Amazon scrapped a resume-screening AI in 2018 because it had learned to penalize resumes mentioning "women's" (e.g., "women's chess club"). Trained on historical hiring data, it inherited historical bias.
3. 🌍 Voice assistants — early Siri and Alexa worked best with American English; non-native accents were misunderstood at higher rates.

How AI bias gets fixed

• Diverse training data (the biggest lever)
• Auditing the AI's outputs across groups
• Human review of high-stakes decisions
• Transparency about what the AI was trained on

Where this comes up in Chippu

Band C (c3-1) is dedicated to AI bias and fairness. Kids learn to spot it in real examples and articulate why it matters.

Related terms

• Training data — the source of most bias
• Machine learning
• Is AI safe for kids?