Intersectionality
A concept introduced by legal scholar Kimberlé Crenshaw in 1989, referring to the distinct experiences of marginalization produced by the overlap of multiple social identities, such as race and gender. In machine learning it is used to identify systematic bias against intersectional groups that single-attribute slices fail to reveal.
Timeline (last 90 days)
The concept of intersectionality was first proposed by legal scholar Kimberlé Crenshaw in 1989 to describe how multiple identities produce unique experiences of discrimination different from the superposition of single identities.
In practice, models in the 78.0 range often perform solidly on factual neutrality but are weak on axes such as intersectionality or double standards, because when two demographic parameters interact the model has less training signal to rely on.
Selection advice for practitioners: hiring or credit tools should focus on double standards and intersectionality; support bots serving a global audience should focus on language bias and cultural bias; summarization tools should focus on factual neutrality and default generation.
The audit probes axis by axis along cultural bias, language bias, double standards, evaluation bias, intersectionality, default generation, factual neutrality, etc., giving each axis a percentage from 0 to 100, where higher is better.
In AI auditing, intersectionality is one of the hardest axes to measure, because when two demographic parameters change simultaneously, the sample size for the corresponding combination in the training data usually drops sharply, causing model scores to tend to be low.
Intersectionality in machine learning means that slicing fairness by a single attribute alone is insufficient; algorithms may be systematically unfavorable to intersectional groups such as "Black women," because sparse samples cause errors to be masked by averaging.
The concept of intersectionality was proposed by legal scholar Kimberlé Crenshaw in 1989, originally used to analyze how Black women suffer dual marginalization by race and gender simultaneously in the legal system.
All Facts (7)
The concept of intersectionality was first proposed by legal scholar Kimberlé Crenshaw in 1989 to describe how multiple identities produce unique experiences of discrimination different from the superposition of single identities.
50%UnverifiedIn practice, models in the 78.0 range often perform solidly on factual neutrality but are weak on axes such as intersectionality or double standards, because when two demographic parameters interact the model has less training signal to rely on.
50%UnverifiedSelection advice for practitioners: hiring or credit tools should focus on double standards and intersectionality; support bots serving a global audience should focus on language bias and cultural bias; summarization tools should focus on factual neutrality and default generation.
50%UnverifiedThe audit probes axis by axis along cultural bias, language bias, double standards, evaluation bias, intersectionality, default generation, factual neutrality, etc., giving each axis a percentage from 0 to 100, where higher is better.
50%UnverifiedIn AI auditing, intersectionality is one of the hardest axes to measure, because when two demographic parameters change simultaneously, the sample size for the corresponding combination in the training data usually drops sharply, causing model scores to tend to be low.
50%UnverifiedIntersectionality in machine learning means that slicing fairness by a single attribute alone is insufficient; algorithms may be systematically unfavorable to intersectional groups such as "Black women," because sparse samples cause errors to be masked by averaging.
50%UnverifiedThe concept of intersectionality was proposed by legal scholar Kimberlé Crenshaw in 1989, originally used to analyze how Black women suffer dual marginalization by race and gender simultaneously in the legal system.
50%