Back to glossary

AI GLOSSARY

Representation Bias

Safety, Alignment & Ethics

A form of algorithmic bias that arises when certain groups, perspectives, or types of data are underrepresented or misrepresented in training data. Representation bias causes models to perform worse for underrepresented groups and can lead to outputs that reflect or reinforce stereotypes.