📢 𝐆𝐨𝐨𝐠𝐥𝐞’𝐬 𝐞𝐱𝐜𝐢𝐭𝐢𝐧𝐠 𝐨𝐧-𝐝𝐞𝐯𝐢𝐜𝐞 𝐀𝐈 𝐮𝐩𝐝𝐚𝐭𝐞𝐬 𝐟𝐫𝐨𝐦 𝐆𝐨𝐨𝐠𝐥𝐞 𝐈/𝐎
At the annual Google I/O developer conference, Google announced a handful of exciting new on-device AI features that will be available in Android phones in the near future:
🤖 𝐆𝐞𝐦𝐢𝐧𝐢 𝐨𝐧 𝐀𝐧𝐝𝐫𝐨𝐢𝐝 𝐟𝐞𝐚𝐭𝐮𝐫𝐞𝐬: It was already known that the “Gemini on Android” digital assistant will be rolled out to millions of devices later this year. Google gave a sneak peak into the exciting features it will include, such as “Ask this video/ PDF” to find specific information in a Youtube video/ PDF, and Gmail drag & drop for AI generated images
📲 𝐅𝐮𝐥𝐥 𝐦𝐮𝐥𝐭𝐢𝐦𝐨𝐝𝐚𝐥 𝐜𝐨𝐦𝐢𝐧𝐠 𝐭𝐨 𝐆𝐞𝐦𝐢𝐧𝐢 𝐍𝐚𝐧𝐨: Google will be introducing Gemini Nano with multimodality, i.e. the ability to understand text, as well as sights, sounds, and spoken language in context, to Android phones, starting with the Google Pixel later this year
🗣️ 𝐂𝐥𝐞𝐚𝐫𝐞𝐫 𝐝𝐞𝐬𝐜𝐫𝐢𝐩𝐭𝐢𝐨𝐧𝐬 𝐰𝐢𝐭𝐡 𝐓𝐚𝐥𝐤𝐛𝐚𝐜𝐤: Gemini Nano’s multimodal capabilities will be leveraged in newer Android versions to improve Talkback, which will provide high quality spoken descriptions of on-screen images (e.g. cut/ style of clothes when shopping online) for users who experience low vision
⚠️ 𝐒𝐜𝐚𝐦 𝐝𝐞𝐭𝐞𝐜𝐭𝐢𝐨𝐧 𝐚𝐥𝐞𝐫𝐭𝐬: Google is also testing a feature that uses Gemini Nano to provide instantaneous alerts if the ongoing phone conversation displays patterns commonly associated with scams (e.g. a “bank rep” asking for sensitive information)
The deployment of several of these features is likely to have been deemed either too privacy-invasive (e.g. scam detection on call) or prohibitively expensive, if done on cloud. This demonstrates the unique value proposition of on-device AI/ ML!
To learn more, read Google’s blog below:
https://lnkd.in/dKpExUGD
If you’re interested in rapidly deploying on-device real-time predictive or generative AI/ ML models in your mobile app, reach out to us at [email protected]
--
1w@