Mark Zuckerberg announced today that Meta, his social media-turned-metaverse-turned-artificial-intelligence conglomerate, is upgrading its AI assistants to give them the voices of a range of celebrities, including Dame Judi Dench and John Cena. But the upgrade that’s more important to Meta’s long-term ambitions is a new feature that lets models see photos and other visual information about users.
Meta today also announced Llama 3.2, the first version of its free AI models with vision capabilities, broadening the usefulness and relevance of AI models for robotics, virtual reality and so-called AI agents. Some versions of Llama 3.2 are also the first optimized to run on mobile devices, enabling developers to create AI-powered apps that run on smartphones and tap the camera, monitor the screen and use the app on the user’s behalf.
“This is our first open source multimodal model, and it will enable a lot of interesting applications that require visual understanding,” Zuckerberg said onstage at the meta Connect event in California today.
Meta has a strong presence across Facebook, Instagram, WhatsApp and Messenger, so the Assistant upgrade allows many people to experience a new generation of AI helper with both voice and visual capabilities for the first time. Meta announced today that more than 180 million people are already using its AI assistant, Meta AI, every week.
Zuckerberg demonstrated several new AI capabilities at Connect. He showed a video of Ray-Ban smart glasses with Llama 3.2 providing recipe suggestions based on ingredients shown on the screen and describing clothes on a rack in a store. The Meta CEO also showed off some experimental AI features the company is working on, including live Spanish-English translation, software that automatically dubs videos into different languages and avatars that answer fan questions on creators’ behalf.
Meta has recently been making AI more prominent in its apps, including building it into the search bars of Instagram and Messenger, and new celebrity voice options available to users include Awkwafina, Keegan-Michael Key, and Kristen Bell.
Meta previously gave its text-based assistant celebrity personas, but those characters didn’t prove popular. The company launched a tool called AI Studio in July to allow users to create chatbots with any persona. Meta said the new voices will be available to users in the U.S., Canada, Australia and New Zealand within the next month. The Meta AI image feature is rolling out in the U.S., but the company didn’t say when the feature will come to other markets.
The new version of Meta AI will also be able to provide feedback and information about your photos — for example, it will tell you what kind of bird you’ve photographed if it doesn’t know what it is — and it will also help you edit your images, adding new backgrounds or details if needed. Google released a similar tool for its Pixel phones and Google Photos in April.
