Be a Fair AI Champion! 🏆

The Shocking Face Recognition Problem

In 2018, MIT researcher Joy Buolamwini tested major facial recognition systems from big companies.

The results were shocking:

| Group | Error Rate | |-------|-----------| | Light-skinned men | 1% | | Dark-skinned women | 35% |

The same AI was 35× more likely to get it wrong for dark-skinned women!

Why? The training data had far more photos of light-skinned faces. AI learned to be great at recognising them — and bad at the others.

🏢 The Amazon Hiring Problem (Again!)

Amazon built an AI to sort job applications. It was trained on 10 years of successful hires — most of whom were men.

The AI learned to penalise resumes that contained words like "women's" (e.g., "women's football team captain").

Amazon scrapped the system when they found out. But it had already been running for years! 😰

Where Does Bias Come From?

Unfair society ↓ Unfair historical data ↓ AI learns unfair patterns ↓ AI makes unfair decisions ↓ At MASSIVE scale 😱 ↓ Makes unfairness worse!

✅ How to Fight AI Bias

1. Diverse training data: Include all groups fairly
2. Diverse teams: People from different backgrounds spot different problems
3. Bias testing: Test AI specifically on underrepresented groups
4. Human oversight: Keep humans in charge of important decisions
5. Transparency: Explain how AI makes decisions

🦸 Safe Habits for Using AI

✅ Always ask: "Could this be wrong?" ✅ Verify medical/legal/safety info from real sources ✅ Tell an adult if AI says something worrying ✅ Don't share personal info ✅ If something feels off, trust your instincts ✅ Take breaks — AI tools are designed to be engaging!

💡 You Are the Future!

The people who will FIX AI bias are:

• Researchers who test AI fairly (like Joy Buolamwini!)
• Engineers who build more diverse training datasets
• Policy makers who write AI fairness laws
• Kids who grow up understanding these issues

That's you. Learning this stuff at your age means you're already ahead! 🚀