Google launches AI model that turns sign language into text on Pixel 11

Google's AI lab DeepMind has launched a large-scale sign language-to-text translation model, bringing sign language AI to consumer products for the first time. It works with American Sign Language initially, with more languages and devices coming later.

The model was trained on 100,000 hours of data covering over 50 sign languages. It converts sign language into pose keypoints instead of sending raw video to servers, protecting user privacy. Unlike simple word-by-word conversion, it uses real machine translation to read spatial information directly.

Pixel 11 phones with Gboard and Live Transcribe now support this feature. About 70 million deaf people worldwide use over 200 sign languages. Tests show signing in ASL is faster and more natural than typing, giving deaf users a more flexible way to communicate.