Gender Shades found error rates up to 34.7% for darker-skinned women versus 0.8% for lighter-skinned men

In “Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification” (Joy Buolamwini and Timnit Gebru, FAT 2018), the authors evaluated commercial gender-classification systems and reported that “darker-skinned females are the most misclassified group (with error rates of up to 34.7%)” while “the maximum error rate for lighter-skinned males is 0.8%.”