Module 2/4 22 min
Bias: where it comes from and how to reduce it
Step 1 / 3·Where bias comes from
0/3 steps done
Learning objectives — after this module you will:
- Understand bias comes from data, design, and use
- Know the protected characteristics that must not drive decisions
- Apply guardrails to reduce bias in HR work
Learn ~6 min
Where bias comes from
Understand the three sources of bias to guard the right places.
Three common sources of bias
- Data: learning from a prejudiced past (like the Amazon case) — AI repeats the past, it doesn't create a fairer future
- Design: choosing variables/signals correlated with protected characteristics (e.g., area code or school name can correlate with gender/origin)
- Use: trusting AI blindly, without oversight, without allowing appeals
Protected characteristics — do NOT use for scoring
Absolutely never use — and don't let AI infer then use — these characteristics: gender, age, ethnicity, religion, marital status, health/disability status, pregnancy. HR decisions may only be based on JOB-RELATED criteria (competence, experience, results). This is the very principle built into TTD's hiring tools.
Key takeaway: Bias comes from data, design, and use; never use protected characteristics to decide.