Meta’s ChatGPT killer is taking over your favorite apps
The AI hype is in full swing right now, transcending form factor boundaries with mixed success. But for the most part, the situation has been a pricey endeavor for users willing to tap into its full potential. Earlier today, Meta AI made its grand debut, drawing power from the Llama 3 model. It’s free (for now) and it could well be one of the first truly mass-market AI products.
That’s primarily because Meta AI is already making its way to some of the most widely used platforms on the planet that draw billions of users. Available in the English language, the multifunctional AI tool can now be found on WhatsApp, Instagram, Facebook, and Messenger.
In its first phase of expansion beyond the U.S. market, Meta AI is making its way to users in Australia, Canada, Ghana, Jamaica, Malawi, New Zealand, Nigeria, Pakistan, Singapore, South Africa, Uganda, Zambia, and Zimbabwe.
Meta has rolled out its AI in the form of a standalone chatbot entity in each of the apps mentioned above. It can even be used in WhatsApp Group chats by simply summoning it using the @MetaAI keyword in the message field. On Instagram and Messenger, it is available in the DMs section.
What can Meta AI do?
Just like a regular chatbot that is tethered to the internet, Meta AI can pull up summarized information from the web. But with Meta’s implementation across its social and messaging apps, it saves users the hassle of a web search. In fact, you can summon it directly in a chat with another person, deploying it as an answering machine of sorts.
Regarding the web search system, it seems Meta AI can pull results from Google Search as well as Microsoft Bing. And just like the aforementioned search engines, it seems the AI assistant will answer users’ questions directly and will also present relevant one-tap search result suggestions in the chat window. The approach is similar to what the likes of Perplexity AI and Arc Search offer.
For example, in a group chat with friends planning a trip, it can find flights and hotel details. It can also create images, just the way any other text-to-image AI tool like Dall-E works. But the most interesting part is that the image changes as you type or alter the description of your prompt in real time.
All you need to do is open a chat with Meta AI and type the command “/imagine” in the message field. To jazz things up, there’s also an “animate” command that brings pictures to life in the form of short animations — or to put it more accurately, shareable GIFs. This is a neat idea that injects some original fun into conversations.
Taking a step further into the realm of social media, Facebook is adding a one-tap “Meta AI” button below public posts, alongside contextual one-tap buttons that directly put the chatbot into action. For example, a post about a mushroom recipe might have buttons that say “safety tips”a nd can directly launch the Meta AI window with a summarized wall of text about the best tips to identify and safely store mushrooms.
The best part about Meta AI’s rollout is that there is no hardware caveat or processing power baseline here. Everything happens on cloud servers and the features are equally accessible in apps that are available all across the world. The only shortcoming is the language barrier, but given Meta’s recent work with large language models, it’s only a matter of time before that problem is solved, as well.
And did I say that it’s all free? Yeah, that too. My only concern is what Meta is getting out of this globally expansive effort. Running generative AI models is one of the costliest endeavors right now. Yes, Meta is not short on cash, but given the company’s shaky record with user data protection and the short supply of quality training material for AI, are these Meta AI interactions going to be the educational fodder for more powerful models?
We don’t have those answers, nor can we predict that. But for now, we have to give it to the company for truly democratizing generative AI in a meaningful and vastly accessible fashion. This all-in-one approach to a multimodal AI tool doesn’t require any subscription or multi-tasking hassle, so there’s that convenience aspect to consider, too.
