AI in the World
Bias, Errors & Limits
6 min+50 XPStep 1/10
Models are mirrors
A model can only learn what's in its data. If historical hiring data shows bias, a model trained on it learns that bias — not because it's evil, but because it faithfully learns patterns.
'Garbage in, garbage out' is an understatement: bias in, bias scaled out — deployed to millions of decisions.
warning
AI doesn't remove human bias — it can launder it into something that looks objective. Always ask: what data trained this?