Understanding AI Bias & Fairness
1. Why does data bias arise even without malicious intent?
Amazon did not intend to discriminate against anyone — the historical data already contained inequality.
2. What is a feedback loop in the context of AI bias?
Predictive Policing: Prediction → more police → more arrests → prediction confirmed.
3. A recruiting tool systematically rates applicants from certain postal codes lower. What is the first step? 2 pts
When bias is active: first stop, then investigate. Do not continue to discriminate while waiting for a solution.
4. What simple method helps to identify bias without being an AI expert?
No expertise needed — only: are there systematic differences between groups?
5. What does EU AI Act Art. 10 require from high-risk systems?
Art. 10 makes bias testing and data management mandatory — not a recommendation.
6. What are warning signs of potential bias? (Multiple possible) 2 pts
Systematic differences, proxy variables, and lack of transparency are warning signals.