⚠️ Risks & rules
Bias
biases
A systematic skew in results, often inherited from one-sided training data or design choices.
💭 Imagine …
If you only read crime novels, the world seems more dangerous than it is. A model mirrors the world of its data.
🔎 Example
An image model shows almost only men for “executive”. A screening model prefers CVs similar to past hires.
🎯 Why it matters
Especially critical when AI assesses people. Check results for fairness and let humans decide.
🎓 Prefer step by step?
The AI Literacy course: 6 short modules with quizzes, a solid foundation in about 60 minutes.