Fairness & bias

Fairness is treating people equitably in how an AI system affects them; bias is a systematic skew in a system's outputs — often inherited from its training data — that works against that goal. ISO/IEC TR 24027 covers techniques for measuring it.