Meta AI Declares War On OpenAI, Google With Standalone Chatbot ...

11 days ago

Forbes

Topline

Meta — the parent company of Instagram, Facebook and WhatsApp — has thrown down the gauntlet on artificial intelligence this week with the release of a standalone AI chatbot and a powerful open source model, Llama 3, positioning itself as a major competitor to bots from the likes of Google and ChatGPT as companies race to develop ever more powerful systems.

Facebook founder Mark Zuckerberg speaks at Georgetown University in a 'Conversation on Free ... [+] Expression" in Washington, DC on October 17, 2019. (Photo by ANDREW CABALLERO-REYNOLDS / AFP) (Photo by ANDREW CABALLERO-REYNOLDS/AFP via Getty Images)

AFP via Getty Images Key Facts

Meta said its AI assistant, Meta AI, is now integrated into the search bar of its Facebook, Instagram, WhatsApp and Messenger apps and can also be accessed when scrolling through the main Facebook feed.

The AI assistant will also be rolled out as a standalone tool through the website Meta.ai, the tech giant said.

While Meta AI has been available within the company’s apps for months, the launch marks the first time it will be available outside of Meta’s social media ecosystem as an independent product.

This restriction has hindered Meta’s ability to compete in the increasingly crowded chatbot market, and by making Meta AI more accessible as a standalone product, the company can now contend more directly with other AI tools, particularly leaders like OpenAI’s ChatGPT, Anthropic’s Claude, Microsoft’s Copilot and Google’s Gemini.

The launch coincided with the launch of Llama 3, the open source AI model Meta used to build the chatbot.

Meta said the Llama 3 models are the “best open models of their class, period” and are potentially capable of besting world-class closed-source AI models from the likes of OpenAI and Google.

Tangent

Meta AI is not universally available around the globe, but the company said it is now rolling the toll out in English “in more than a dozen countries outside of the U.S.” The company said people will now have access to Meta AI in English in Australia, Canada, Ghana, Jamaica, Malawi, New Zealand, Nigeria, Pakistan, Singapore, South Africa, Uganda, Zambia and Zimbabwe. More countries are likely to come, Meta suggested, adding “we’re just getting started.”

Key Background

Since ChatGPT ignited a firestorm of interest in the AI assistant space, the tech industry has been racing to build, develop and release ever more powerful AI models. The models, such as Meta’s Llama, OpenAI’s GPT and Google’s Gemini, are the underlying architecture powering tools like chatbots, image generators and video generators. Besides Meta, multiple firms — including Google, OpenAI and Mistral — have all released new versions of their leading systems within the past month. The decision to make Llama 3 open source marks a notable shift from other market leaders like OpenAI and Google, which tend to keep their technology closed. Open source tech can be used, scrutinized and shared more readily within the industry — though open source licenses can still come with conditions for use — and it often forms a scaffold upon which others build an array of apps and tools. The release of a powerful open source AI model like Llama 3 could fuel such a flurry of innovation designed around Meta’s model. It is not without cost, however, and Meta CEO Mark Zuckerberg has previously said the company is spending billions to develop its AI systems.

What To Watch For

The summer is expected to herald even more AI releases from the industry as players prepare for the expected launch of GPT-5, the next major update to the system powering OpenAI’s ChatGPT. Meta is already training a more powerful AI model to succeed Llama 3, according to the company’s chief AI scientist, Yann LeCun.