foundations

Bias (AI)

Systematic errors in a model's outputs that consistently favor or disfavor certain groups, topics, or responses — often inherited from training data or reinforced during fine-tuning.