AI Bias

AI bias is systematic error in an AI system's outputs that produces unfair treatment of individuals or groups, typically arising from biased training data, biased labels, or model architecture choices that proxy for protected attributes. Common types include historical bias, measurement bias, aggregation bias, and deployment-context mismatch.

Bias detection requires disaggregated evaluation across demographic subgroups — aggregate accuracy metrics mask disparate impact. Mitigation strategies include data rebalancing, fairness constraints during training, post-hoc calibration, and ongoing monitoring after deployment. Regulators in lending, employment, and healthcare require documented bias testing as a condition of AI system approval.