Glossary · Safety

Bias in AI

Systematic skew in a model's outputs that reflects imbalances or prejudices present in its training data, which can disproportionately disadvantage specific groups if not identified and mitigated.