Google spoke to Reuters this week about its plan to develop an industry-standard method to classify skin tones. Many tech researchers and dermatologists are saying that the present classification system is not adequate for determining whether products are biased against people of color.
The Fitzpatrick Skin Type (FST) six-colour scale, which dermatologists have used since the 1970s, is what tech companies use to categorize people and measure whether products like facial recognition software and smartwatch heart-rate sensors work the same for all skin tones. The FST system disregards diversity among people of color, according to critics.
Abandoning the FST
The FST includes four categories for white skin types and one type each for ‘black’ and ‘brown’ color. Researchers at the US Department of Homeland Security said in a federal technology standards conference last October that the FST should be abandoned, citing its poor representation of color range in diverse populations.
In response to a Reuters question about FST, Google said that it has been quietly working on seeking out better measures. However, it declined to offer details about what its effort entails, only calling them “alternative, more inclusive” measures.
A clear and present problem
The controversy about skin tone is part of a larger reckoning with diversity and racism in the tech industry, where there are more white people than in other sectors like finance. There is also the matter of ensuring that technology serves all skin colours, ages, and genders the same.
The matter is now very crucial, given that artificial intelligence, which is increasingly powering most of them, is implemented in systems that govern sensitive and regulated industries like law enforcement and healthcare. Companies know they can have faulty products for people under-represented in research and testing data, and for Google, this is one way to prevent that.