The Gender Shades Project exposed skin type and gender bias in commercial AI products from IBM, Microsoft, Amazon and more, galvanizing research and public attention on issues of algorithmic bias.
Articles across 40 countries
“When I started graduate school at MIT, I endeavoured to write a master's thesis that might create a new field of inquiry or perhaps awaken the world to issues largely hidden. I succeeded beyond my wildest dream with the Gender Shades project. My 2017 thesis showed the prevalence of pale male face datasets, introduced the concept of intersectional phenotypic and demographic analysis, produced the Pilot Parliaments Benchmark and formed the basis of two peer-reviewed publications.“