Bias in AI
Domain: Safety & Ethics
🎯 Motivation
AI systems can reflect and amplify human biases - racial, gender, cultural. Bias detection and mitigation are essential.
📋 Concrete Example
A hiring AI that favors male names over female names is biased. Cause: training data had more male hires. Fix: balanced training data, fairness metrics, regular audits.
🔗 Analogy
A judge who is stricter with people from certain backgrounds - that is bias. AI bias is the same: the system treats some groups unfairly. Detecting and fixing bias is essential.