Site icon CrypLogger.com

Google has developed a scale of skin tones to combat “biased” AI

Google has opened the source code of the Monk Skin Tone (MST) 10-point skin tone scale to open-ended AI researchers and developers. In the coming months, the company plans to implement it in a number of its products.

During the annual developer conference I/O 2022, the company showed how scale can improve Google Search. For example, when searching for cosmetics, users can refine their skin tone and get more relevant pictures.

Image search for “Wedding makeup” with the ability to specify skin tone. Data: Google.

The company is working on a standardized way to mark up web content that will allow creators and brands to mark content with attributes like skin tone, hair color and texture.

The MST scale will also appear in Google Photos. The company will update the Real Tone artificial intelligence photo post-processing feature for Pixel phones introduced in 2021. According to the developers, the MST scale will improve the display of faces with all shades.

New Real Tone filters coming to Google Photos on all available platforms in the coming weeks.

MST was created in collaboration with Harvard professor Ellis Monk. It is designed to cover a wider range of skin tones.

The company also said the scale will help create more representative datasets for training and testing AI models for bias.

During I/O 2022, Google introduced a number of other products and features for its AI services.

Recall that in April, the developers of Google Lens added a simultaneous search by image and text.

In October 2021, the tech giant introduced Pixel smartphones with a tensor processor for machine learning.

Subscribe to Cryplogger news in Telegram: Cryplogger AI – all the news from the world of AI!

Found a mistake in the text? Select it and press CTRL+ENTER

Exit mobile version