Representation Bias

Bias arising when certain groups or perspectives are underrepresented or overrepresented in training data, leading to models that perform poorly for underrepresented populations.

When some groups are underrepresented in training data. If a facial recognition AI is trained mostly on light-skinned faces, it will work poorly on darker-skinned faces.

Representation bias is a governance priority because it directly causes AI systems to underperform for underrepresented groups. Your data governance standards should require representation analysis before AI training data is approved for use.