Meta recently introduced its AI chatbot, driven by Llama 3, into the search bar across all of its applications.
Meta is making significant strides today to showcase its AI offerings throughout its platform. They’ve rolled out an upgrade to their AI chatbot, now powered by their latest large language model, Llama 3, and it’s live in the search bars of their four major apps (Facebook, Messenger, Instagram, and WhatsApp) across various countries. Alongside this update, they’ve introduced other new features like quicker image generation and access to web search results.
This development confirms and expands upon a recent test that TechCrunch covered last week, where they noticed Meta AI being tested on Instagram’s search bar.
Furthermore, Meta is launching a new meta.ai website, providing users with direct access to the chatbot.
The news highlights Meta’s efforts to establish itself as a significant player amidst the current enthusiasm for generative AI tools among consumers. In a bid to rival other popular services like those from OpenAI, Mark Zuckerberg asserted today that Meta AI could potentially be the “most intelligent AI assistant that you can freely use.”
Meta initially introduced Meta AI in the U.S. last year and is now broadening the chatbot’s availability in the English language to over a dozen countries, including Australia, Canada, Ghana, Jamaica, Malawi, New Zealand, Nigeria, Pakistan, Singapore, South Africa, Uganda, Zambia, and Zimbabwe.
Last week, the company commenced testing Meta AI in countries such as India and Nigeria; however, notably, India was absent from today’s announcement. Meta stated that it intends to keep Meta AI in test mode in India for the time being.
“We continue to learn from our user tests in India. As we do with many of our AI products and features, we publicly test them in various phases and in a limited capacity,” stated a company spokesperson.
New features
Previously, users could already inquire with Meta AI for writing or recipe suggestions. Now, they have the added capability to request web-related results, powered by both Google and Bing.
The company has also announced enhancements to image generation, aiming to accelerate the process. Additionally, users now have the option to request Meta AI to animate an image or convert it into a GIF. What’s more, users can witness the AI tool in action, modifying the image in real-time as they input their commands. Furthermore, the company has made strides in improving the image quality of AI-generated photos.
Meta asserts that its new model has exhibited improvements in addressing a longstanding issue with AI-powered image-generation tools: spelling accuracy.
All AI things everywhere at once
Meta is taking a comprehensive approach by ensuring that Meta AI is accessible across various platforms. The company is making the bot available not only in the search bar but also within individual and group chats, as well as in the feed.
According to the company, users can now pose questions related to posts in their Facebook feed. For instance, if they come across a photo of the aurora borealis, they could inquire with Meta AI about the optimal time to visit Iceland to witness the northern lights.
Meta AI is already integrated into Ray-Ban smart glasses, and the company has announced plans to make it available on the Meta Quest headset in the near future.
However, there are drawbacks to deploying AI across numerous platforms. One notable concern is the potential for models to generate nonsensical or inappropriate responses, a phenomenon sometimes referred to as “hallucinating.” This widespread deployment could pose challenges for content moderation.
For instance, earlier this week, 404 Media reported an incident where Meta AI, interacting in a parents group, provided an unusual response, claiming to have a gifted and academically challenged child attending a specific school in New York. The odd message was noticed by parents, prompting Meta to intervene and remove the response. They acknowledged the issue and committed to ongoing improvements.
In response to this incident, Meta stated, “We share information within the features themselves to help people understand that AI might return inaccurate or inappropriate outputs. Since we launched, we’ve constantly released updates and improvements to our models, and we’re continuing to work on making them better.”
