Google is developing a new way of classifying skin tone and color on their platforms, including emojis, which recently received new handshake shades, as you can see see here. The intention of the new format is to avoid possible accusations of bias and artificial intelligence errors, something for which the company has been heavily criticized.

The problem lies precisely in the Fitzpatrick Scale, used since the 1970s and which classifies skin into four shades of white and two shades of dark. This scale was originally developed to classify skin types according to their tendency to suffer from sunburn and was used to apply ultraviolet ray treatments.

Over time, this same model began to be adopted by companies that wanted to classify groups of people. In addition to emojis from Google, this skin color scale is also used by artificial intelligence systems to recognize users.

Change in Google emoji skin tone scale

However, this scale has come to be considered error-prone due to the limited dark skin tones present in the model. This has already occurred with smartwatches and heart rate measurement methods, which present more errors on dark skin. In addition, facial identification in Google Photos, for example, has already shown to have problems with this.

“We are working on alternative, more inclusive measures that may be helpful in developing our products and will collaborate with scientific and medical experts, as well as groups working with communities of color,” Google told Reuters.

It is not yet clear what new system Google will adopt to classify skin tone in emojis and machine learning systems. But the changes are likely to be made quietly, and include a wider variety of tones.

Through which channels you reach those people, classic and out of the box. Reuters