Google will adapt its search engine to the skin color of Internet users

The company presented a palette of ten skin tones that will allow a better representation of search results, especially on Google Images.

On the occasion of its 2022 annual conference, Google presented its latest projects and products to be released. Among these, the company announced a new tool to strengthen the representativeness of different skin colors on the web.

Specifically, it is a new spectrum of ten skin colors, created in partnership with doctor and sociologist Ellis Monk. This tool is called the Monk Skin Tone Scale, repository of Monk skin colors, in English.

This palette will allow, for example, artificial intelligence to better recognize faces, regardless of their skin tone. Algorithms, from Google or other tech giants, are regularly criticized for their ignorance of black skins, opening the door to discrimination bias.

These tools generally have difficulty recognizing or identifying people with darker skin. Several companies have also become aware of this in recent years, and have announced improvement projects.

The 10 shades of the Monk Skin Tone Scale
The 10 shades of the Monk Skin Tone Scale © Google

Improved search results

In fact, a palette of different skin colors will be offered under the search engine in order to direct the user to the most relevant results.

“When you search for example makeup-related images in Google Images, you will now see an option to narrow down your results by skin tone. So if you search for ‘everyday eyeshadow’ or ‘wedding makeup “, you will find results better suited to your needs”, specifies the press release from Google.

The Improved Google Search Tool
The improved Google search tool © Google

Finally, the palette will also be integrated with the automatic corrector of Google Photos, in order to better reflect the original hues.

Google has given it a collaborative vocation by encouraging Internet users to use “this new inclusive schema to label their content with attributes such as skin tone, color and hair texture”. In this way, their content should be better referenced by other platforms.

As early as last year, Google revealed an improvement in its facial recognition methodologies, in particular by abandoning the historic Fitzpatrick classification, created in the 1975s and considered too little invented. The latter includes six skin tones, four of which are white and only two dark.

Leave a Comment