AI Bias

Ethics and Safety

When an AI system produces unfair results for certain groups because of imbalances or prejudices baked into its training data.

AI learns from data produced by people, and that data carries human history and prejudice. A hiring model trained on past decisions can learn to prefer the same profiles that were preferred before. A face recognizer trained mostly on some demographics performs worse on others.Bias is rarely intentional but always consequential, especially when AI helps decide loans, jobs, or policing. Testing for it and correcting it is a core part of responsible AI work.