Meta Tests AI Chatbot On WhatsApp, Instagram, Messenger

Meta has confirmed to TechCrunch that it is testing Meta AI, its large language model-powered chatbot, with WhatsApp, Instagram and Messenger users in India and parts of Africa. The move signals how Meta plans to tap massive user bases across its various apps to scale its AI offerings.

The social media giant has been scrambling to roll out more AI services in the wake of big AI moves from other major tech companies, OpenAI and more.

Meta announced plans to build and experiment with chatbots and other AI tools in February 2023. India, where users have recently started noting the appearing of the Meta AI chatbot, is a very important market for the company: it is home to more than 500 million Facebook and WhatsApp users, making it Meta’s largest single market.

Developing markets, where smartphone users are growing faster than developed markets like the U.S. (where growth has plateaued), are also a big target for Facebook to try out more services to engage audiences. Users in Africa are also reporting signs of Meta AI appearing in WhatsApp.

Meta confirmed the move in a statement. “Our generative AI-powered experiences are under development in varying phases, and we’re testing a range of them publicly in a limited capacity,” a Meta spokesperson told TechCrunch.

Meta unveiled Meta AI, its general-purpose assistant, in September 2023. The AI chatbot is designed to answer user queries directly within chats as well as offer them the ability to generate photorealistic images from text prompts. In the case of Instagram, there’s evidence it’s also being used for search queries.

Meta has been somewhat late to the game for building and rolling out AI tools to its users. In part, its teams assumed that generative AI tech was not quite ready for prime time. OpenAI clearly proved that wrong, putting MetaAI on the back foot.

“The availability of ChatGPT somehow captured the attention and enthusiasm of the public,” said Yann LeCun, the Turing Award winner who is Meta’s chief AI scientist, speaking earlier this week at an “AI Day” that the company organized at its offices in London. “What was surprising to people like me about ChatGPT was not the technology or the performance of the system. It was how much interest it gathered from the public. That surprised everyone. It surprised OpenAI, too.” Meta, he explained, thought that AI chatbots, based on its own efforts to launch them, “were not particularly welcome… in fact, some of them were trashed by people.” Now, he described the company, and the wider tech community, as “more open, and more comfortable with releasing models.”

And that’s what Meta is doing now. More pragmatically speaking, though, there are three reasons why Meta may be forging ahead with its AI strategy.

First, for user retention (users now expect to see and want to use AI tools in their apps; if Meta doesn’t offer them the worry is that those users will move away).

Second, for investor retention (investors want strong earnings, sure, but in tech they also want to see signs that Meta is committed to backing in and building what many believe will be the next generation of computing).

Third, for its own pride (it’s been setting the pace for so much in areas like mobile apps, social media and advertising for the last decade, and it has outsized talent on its bench, including the celebrated AI academic Yann LeCun. Is it really going to jump the shark and miss all of this?!).

Instagram and WhatsApp’s massive global user base, boasting billions of monthly active users, to be sure presents Meta with a very unique opportunity to scale its AI offerings. By integrating Meta AI into WhatsApp and Instagram, the Facebook-parent firm can expose its advanced language model and image generation capabilities to an enormous audience, potentially dwarfing the reach of its competitors — at least on paper.

The company separately confirmed earlier this week that it will be launching Llama 3, the next version of its open source large language model, within the next month.