Meta increases pressure on OpenAI and Google with free speech AI

Hamburg The US technology group Meta Platforms has released the new version of its artificial intelligence (AI) Llama – and made it available to business customers free of charge. Start-ups and other companies can now tailor the language model to their needs or use it to develop their own services, the company said on Tuesday evening.

Experts expect that the already rapid further development of and competition among the individual AI systems will subsequently gain even more speed. “We take a different, decentralized approach to AI,” said Nick Clegg, President of Global Affairs at Meta, the Handelsblatt. Before joining Meta, Clegg was leader of the Liberal Democrats and also for a time Deputy Prime Minister of Great Britain.

Meta does not want to build the one “singing, dancing chatbot for all occasions”, but rather spread AI capabilities across its existing product range. It therefore makes sense to make the basic technology available free of charge, as it can be improved more quickly. According to Clegg, Meta already sees itself at the forefront of the tech industry when it comes to the use of AI technology. “We’re investing more than $10 billion this year alone.”

With the release of Llama 2, the Facebook parent company is also trying to catch up on its development deficit and put pressure on important competitors such as OpenAI or Google. Because another, desired effect should be the price pressure that Meta creates with the free publication. For example, OpenAI’s paid GPT-4 language model is superior to Meta’s, but Llama is quite competitive.

Meta Manager Nick Clegg

According to Clegg, Meta wants to release commercial chatbots “relatively soon”.

(Photo: AFP/Getty Images)

Meta released a first version of Llama earlier this year and made it available to AI experts and researchers for experiments. According to a company announcement, Microsoft of all people, which is the godfather and most important investor behind the market leader OpenAI, is now supposed to help with the further spread. Microsoft is “our preferred partner to commercialize Llama 2,” it said. Meta founder and CEO Mark Zuckerberg even published a photo on Tuesday evening showing him together with Microsoft boss Satya Nadella. Zuckerberg thanked him for the good cooperation.

Meta and Microsoft want to “democratize” AI

In a blog post published at the same time, Microsoft formulated the “common commitment” of both companies to “democratize” AI technology. Llama will also be adapted for the popular Windows operating system, so that it can also be used on private computers.

The software company is known for deliberately allowing such ambivalences. Thanks to his participation in the ChatGPT developer OpenAI, whose systems are interwoven with Microsoft’s Office programs, he currently holds a dominant position in the AI ​​segment anyway. By partnering with Meta, Nadella reduces its dependency and at the same time strengthens the attractiveness of Microsoft’s cloud service Azure. There, Llama is made available to its own customers, from which the group earns indirectly.

However, the partnership between Zuckerberg and Nadella is not exclusive: Meta’s AI is available to customers of Azure as well as those of the market leader Amazon Web Services (AWS) and other providers. The chip manufacturer Qualcomm also announced that it would also like to make Llama available on laptops and smartphones. So the AI ​​could also be used without cloud access.

>> Read about this: How Amazon wants to catch up in the AI ​​race

Llama is a so-called generative language model. It can independently generate text, images or program code. Applications such as chatbots or search engines can be based on it. OpenAI’s ChatGPT or Google’s Bard are based on similar systems.

Meta’s open approach, referred to as “open source” in industry jargon, distinguishes the group from its competitors, who protect the program code of their models – and are already making money today.

According to a report by the “Financial Times” from mid-July, Meta was at least considering asking companies to pay for the individual design of a Llama-based service. Apparently, these were initially rejected.

Satya Nadella (left) and Mark Zuckerberg

Zuckerberg thanked for the good cooperation.

(Photo: Instagram)

However, a limitation applies to services that count more than 700 million monthly active users. Competitors such as the social network Snapchat or the news service Telegram can only use Llama if they negotiate an individual license with Meta.

Read more about artificial intelligence

According to Meta, with its generosity it wants to promote the further development of AI tools and reduce fears of the technology. However, the open source approach also offers the opportunity to quickly build up a large user base and a growing repertoire of additional services despite the backlog. In order to improve generative AI, large amounts of data are required.

Meta is “not a prosperity organization”

In addition, the openness should fuel the further development of Llama. “We’re a company, not a wealth organization,” said Nick Clegg. His group hopes to be able to integrate innovations from external developers into its own products in the future.

The procedure then probably becomes a business indirectly. The group plans to offer AI chatbots for the social network Instagram or the short message service WhatsApp. Companies could use these to communicate with customers, for example. The power of the bots would increase with each development step of Llama. According to Clegg, the offers can be expected “relatively soon”.

“There is a future,” Meta’s AI boss Yann LeCun recently prophesied in Paris, “in which each of us will communicate via intelligent agents most of the time.”

Meta’s core business – advertising that is as targeted as possible – is already being driven forward with the help of its own AI systems.

More: Amazon is planning offers for Internet and mobile communications

source site-13