Meta heats up the AI chatbot race with new assistant powered by Llama 3

Meta AI will be available across the search bar of its apps such as Facebook, Instagram, WhatsApp and Messenger in more than a dozen countries including the United States, Australia, Canada, Ghana, Jamaica, Malawi, New Zealand, Nigeria, Pakistan, Singapore, South Africa, Uganda, Zambia and Zimbabwe.

Users will also be able to access the chatbot on the web through a new meta.ai website that debuted today.

Meta AI's India test

India, which is Meta's largest market in terms of user base with a combined userbase of over a billion users, is however notably missing from the first wave of countries that will have access to the new chatbot.

Last week, Moneycontrol reported that the social networking giant was testing Meta AI with select users in India across its suite of products, including Instagram, Messenger and WhatsApp. It appears that the country will now have a longer test period.

"We continue to learn from our user tests in India. As we do with many of our AI products and features, we test them publicly in varying phases and in a limited capacity," a Meta spokesperson said in a statement.

Meta first introduced Meta AI in September 2023 as part of the social networking giant's big push into generative AI.

Users can interact directly with the chatbot, which functions as a general-purpose assistant capable of answering a wide range of questions, including real-time information powered by Google and Bing.

One can also generate text and images, summarise long pieces of text, help with writing tasks such as proofreading, editing, translate text from one language to another, create poems and stories among others.

People can also call in the assistant in their existing personal and group chats on Instagram to get advise or ask questions. For instance, one can ask for recommendations for a group trip, or ideas for dinner party recipes, following which it will offer options directly in the chat. One can type “@” and then tap Meta AI to converse with the chatbot.

Real-time image generation

Meta said that it is also making image generation faster, so that people can create images from text in real-time using Meta AI’s Imagine feature.

Users can see an image appear as they start typing and it will change with every few letters typed, the company said. This feature is rolling out today in beta on WhatsApp and the Meta AI web experience in the United States.

Meta AI’s Imagine feature allows users to create images from text in real-time

In addition, users can also ask the chatbot to animate an image, iterate on it in a new style or turn it into a GIF to share with their friends.

Meta said that it has worked on improving the quality of the images generated, which are now sharper and higher quality, with a better ability to include text in images.

"From album artwork, to wedding signage, birthday decor and outfit inspo, Meta AI can generate images that bring your vision to life faster and better than ever before" the company said in a blogpost.

Facebook users will also be able to access Meta AI while they are scrolling through the app's main feed. If they come across any post they are interested in, one can ask Meta AI for more information about it.

For example, if one sees a photo of the northern lights in Iceland, they can ask Meta AI what is the best time of this year to check out the aurora borealis.

Llama 3 launch

This launch coincided with the social networking giant releasing Llama 3, the latest and the most powerful version of its open source large language model with new capabilities including improved reasoning.

The company has released two models - one with 8 billion parameters and another with 70 billion parameters that can support a broad range of use cases.

Zuckerberg said they plan to release more models in the future that will bring in capabilities such as longer context windows, multilingual, and multi-modal, which means it can understand and work with different formats, including text, code, audio, image, and video, at the same time.

He also said the company is also training a larger dense model with over 400 billion parameters.

"With Llama 3, we set out to build the best open models that are on par with the best proprietary models available today. We are embracing the open source ethos of releasing early and often to enable the community to get access to these models while they are still in development" the company said in a blogpost.