Google DeepMind introduces sign-language-to-text (SL2T), a breakthrough AI model powering new sign language features for Deaf and hard of hearing users. SL2T enables sign-to-text dictation in Gboard and Live Transcribe on Pixel 11, starting with American Sign Language (ASL) to English, with more devices and languages coming soon. This brings sign language AI into consumer products for the first time, allowing Deaf users to sign to their phone anywhere they'd normally type.

SL2T is trained on over 100,000 hours of data across more than 50 sign languages. It sees sign language as pose landmark locations rather than raw video, protecting user privacy. The model achieves a remarkable zero-shot score of 70 BLEURT on the FLEURS-ASL benchmark, significantly higher than any previously reported score.

The technology was developed with extensive input from the Deaf community, including Sam Sepah, a Deaf Googler, and the AI Sign Language Advisory Committee (AISLAC). Google DeepMind emphasizes building with the community, not just for it, ensuring responsible real-world deployment.