- Meta AI is now powered by the newly released Llama 3.2 multimodal models on WhatsApp, Instagram, Facebook, and Messenger.
- Thanks to its multimodal nature, users can now converse with Meta AI using voice and produce AI-generated images of themselves.
- Meta AI now lets you upload images for AI photo editing and you can also ask questions about the image.
With the release of new Llama 3.2 multimodal models —11B and 90B — Meta has unlocked new use cases for its Meta AI chatbot. At the Meta Connect 2024 event, the company announced several new features for Meta AI that allow users to interact with various modalities like audio and images.
Talk to Meta AI
First and foremost, you can now talk to Meta AI using voice and it will reply out loud. You can continue the conversation and ask questions on any topic. The best part is that it can find even the latest information by browsing the internet.
It’s not as conversational as Gemini Live and ChatGPT Advanced Voice, but you get a standard two-way voice chat interface. There is no support for interruptions, though.
The timing of this announcement couldn’t be any better as ChatGPT Advanced Voice Mode started rolling out to users today. Meta Voice chat is available through the Meta AI chatbot on WhatsApp, Facebook, Messenger, and Instagram DM. There are different AI voices available and you can even choose the voice of public figures such as John Cena, Keegan Michael Key, Awkwafina, Dame Judi Dench, and Kristen Bell.
Image Analysis
Since Meta AI is now powered by Llama 3.2 11B and 90B multimodal models, you can upload an image and ask Meta AI to analyze it. For instance, you can upload an image of a mountain, ask where it is located, and find more information along the way. You can also choose to upload charts and diagrams and infer meaning from your visual input.
AI Photo Editing
Next, Meta AI brings AI photo editing to its social media apps. You can upload an image and ask the AI chatbot to change the background, erase unwanted objects, change outfits, and much more.
Basically, AI photo editing is now readily available on Meta’s social stack, including WhatsApp, FB Messenger, Instagram, and Facebook. It works similarly to Google’s Magic Editor, but it’s available within your social media apps and you can seamlessly share them as stories.
Create AI Images of Yourself
Best of all, the Reimagine AI tool now lets you create AI-generated images of yourself. You can reimagine your photos from feed, stories, and Facebook profile pictures by simply adding a prompt, and Meta AI will instantly generate an image based on your prompt. It means that you don’t have to train your images using LoRA to create AI-generated images of yourself.
My colleague and boss Devinder is in Palo Alto attending Meta Connect 2024. He got a chance to go hands-on with this new Meta AI capability in WhatsApp and generate some cool photos.
AI Translation of Reels
Last but not least, one of the promising features of Meta AI is the automatic translation of Reels. If a creator has published a Reel in a foreign language that you don’t understand, Meta AI will translate the audio into your language automatically with perfect lip-syncing. Currently, the feature is limited to Latin America and the US in English and Spanish. Meta says the feature will be expanded to more regions and languages pretty soon.
Next, Facebook and Instagram users may see Meta AI-generated images in the user feed based on user interest or current trends. You can also tweak the prompt to generate new content in your feed. And finally, users will be able to personalize themes using AI in their private DMs.
So these are the new Meta AI features coming to WhatsApp, Instagram, Facebook, and Messenger. Are you excited to check them out? Let us know in the comments below.
