Google’s new skin tone scale could reduce AI’s racial bias

We’ve already seen some cases of AI being biased towards lighter skin tones. But Google wants to turn it around and it’s now using an alternative, more inclusive skin tone scale.

Teaming up with a Harvard professor, Google is using and open-sourcing the Monk Skin Tone (MST) Scale. It’s more inclusive than the current tech-industry standard, making various skin tones included in search results.

Dr. Ellis Monk is the person behind the MST Scale. It’s a 10-shade scale that “provides a broader spectrum of skin tones that can be leveraged to evaluate datasets and ML models for better representation,” as Google explains.

“In our research, we found that a lot of the time people feel they’re lumped into racial categories, but there’s all this heterogeneity with ethnic and racial categories,” Dr. Monk says. “And many methods of categorization, including past skin tone scales, don’t pay attention to this diversity. That’s where a lack of representation can happen…we need to fine-tune the way we measure things, so people feel represented.”

Not only is the scale more accurate in representation of all skin tones, but Google has also decided to openly release it to the machine learning community. “By openly releasing the scale to the broader industry, we hope others will incorporate the scale into their development processes and that we can collectively improve this area of AI,” Google writes.

Google itself was at the center of a scandal a few years ago when its algorithm tagged a couple of African Americans as “gorilla.” Google didn’t exactly work out the solution – they just removed the “gorilla” label altogether, which I think is far from resolving the issue.

I guess now it’s time to make up for past mistakes. The MST Scale should help not only Google but other AI-based tools to recognize skin tones with more accuracy and broader representation.

You can access the scale at skintone.google, and read more about it on Google’s blog.

[via The Verge]