EmbeddingGemma 2: an open, lightweight multimodal embedding model
deepmind.google·6d ago
AI Engineering
DeepMind Blog·Aug 12·5 min read
Putting sign language AI into users’ hands
TL;DR
DeepMind has developed a sign language AI, SL2T, that integrates American Sign Language input into Gboard and Live Transcribe, emphasizing collaboration with the Deaf community. This initiative aims to enhance accessibility and expand into more sign languages and AI capabilities.
✦ Why It Matters
Explore SL2T in Gboard and Live Transcribe to enhance communication accessibility today.
Key Takeaways
- Test SL2T on Pixel 11 for ASL input.
- Engage with the Deaf community for feedback.
- Monitor updates for additional sign language support.
- Consider implications of AI in accessibility projects.
Full Summary
DeepMind's SL2T (Sign Language to Text) model allows users to input American Sign Language (ASL) on their mobile devices through Gboard and Live Transcribe. The project was shaped by the Deaf community, involving Deaf individuals in every phase from conceptualization to testing.
An AI Sign Language Advisory Committee (AISLAC) was formed to ensure that the technology aligns with the needs of the Deaf community, resulting in a joint impact report that outlines the technology's capabilities and limitations. SL2T is currently available on Pixel 11 devices, with plans for broader device compatibility.
The initiative is part of Google's mission to make information universally accessible, aiming for parity with spoken languages. Future developments will include support for additional sign languages and advanced AI features.