On July 13, according to media reports, Meta is preparing to release a commercial version of its artificial intelligence model, allowing startups and enterprises to build custom software based on the technology.。

The move will allow Meta to compete with Microsoft-backed OpenAI and Google, which are currently well ahead in the race to develop generative AI.。Generative AI can create text, images, and code, and is powered by large language models (LLMs), which are trained on large amounts of data and require powerful computing power.。

   

LLM open source, Meta new model details will be made public

  

Meta released its own language model, LLaMA, to researchers and academics earlier this year.。The upcoming new version will be more widely used and customized by the public。

On top of that, Meta has said its LLM is open source, meaning details of the new model will be publicly released.。This is in stark contrast to what competitors like OpenAI are doing。

OpenAI did not disclose the data and code used to build the model in its latest model, GPT-4, a practice that has been criticized.。Regulators also disagree。The EU's regulators mentioned some time ago in the development of the relevant AI regulation draft that companies deploying generative AI tools such as ChatGPT will have to disclose any copyrighted material used to develop their systems.。Sam Altman, co-founder and CEO of OpenAI, said: "The current draft EU AI law is over-regulated.。"

YannLe Cun, vice president and chief artificial intelligence scientist at Meta, said in a meeting last week: "The competitive landscape of artificial intelligence will completely change in the next few months or even the next few weeks, when there will be open source platforms, and they will be as good as non-open source platforms.。"

The benefits of the open source model include higher user usage, which allows more data to be input for AI processing。The more data an LLM has, the more powerful it will be。In addition, open source models allow researchers and developers to find and resolve bugs while improving technology and security.。This is the "short board" of many tech companies, including Meta, which for years have faced scrutiny from regulators for their various privacy and misinformation.。

Meta's upcoming commercial release of AI comes as Silicon Valley technology groups such as Microsoft and Google compete fiercely for AI dominance。

Nick Clegg, Meta's head of global affairs, expressed his appreciation for "open source" in the media, saying that "openness is the best antidote to the fears surrounding artificial intelligence."。The move will also help Meta catch up with competitors, as the open model will allow companies of all sizes to improve technology and build applications on top of it.。

Meta has been working on AI research and development for more than a decade, but since OpenAI launched ChatGP, Meta seems to be in a bad situation and has become more passive, as the emergence of ChatGPT has prompted other large technology companies to launch similar products.。

One person with knowledge of Meta's senior strategy said, "Meta's goal is to weaken OpenAI's current dominance.。"

  

There is no charging plan for the commercial version of AI model

   

While Meta's technology is open source and currently free, people familiar with the matter said the company has been exploring charging corporate customers to use their own proprietary data to fine-tune the model for their own needs.。One person said there are currently no plans to charge, and Meta will not do so in an upcoming release.。

Joelle Pineau, vice president of AI research at Meta, declined to comment on the development of the new version of the AI model and how it could be commercialized, but said that while some open-source stuff has been released, it doesn't indicate a complete abandonment of the relevant intellectual property.。He added: "We are not ashamed to admit that we do want to use these models in our products.。"

In 2021, Meta CEO Mark Zuckerberg announced that he will focus on creating a digital world full of avatars, the metasurface, and spend more than $10 billion a year on the project.。But in the end it turns out that this costly investment is not popular with investors。So after the emergence of ChatGPT, Meta is no longer only focused on the meta universe, but chose to actively embrace AI, and announced to increase investment in artificial intelligence。

Earlier this year, Meta launched a generative artificial intelligence division, led by Chris Cox, Meta's chief product officer.。According to Meta, Cox's team covers AI research and related product development as Meta is creating entirely new businesses.。

Zuckerberg and other executives have hinted that they will create multiple AI chatbots for individuals, advertisers and businesses through Meta social media platforms such as Instagram, WhatsApp and Facebook, with the support of their LLM.。

While providing software for free may not seem to have anything to do with profitability, experts believe that companies can also use this strategy to capture market share。

A person familiar with the matter said that Meta realized that they were behind in the current AI race, so they chose open ecosystems (open source) as their new strategy.。At the moment it looks like Meta is doing the right thing。

   

Open source is a double-edged sword

  

While open source is good for businesses, for now, open source AI still has a clear risk of being shaped and abused by people with ulterior motives.。For example, child safety groups report an increase in current images of child sexual abuse generated online by artificial intelligence.。

The researchers also found that the AI model "BlenderBot 2," released in 2021, is spreading misinformation.。"BlenderBot 3" will improve this, says Meta。But now there are still users found it generates some false information。

In addition, there are regulatory and legal risks in intellectual property and copyright。On July 10, comedian Sarah Silverman filed a lawsuit against Meta and OpenAI, saying their work was used to train AI models without her consent.。

Meta released its AI open source model LLaMA to researchers in February。A month later, LLaMA was widely leaked by online forum 4chan, resulting in commercial development by developers on its basis, in violation of Meta's licensing rules, which state that it is not allowed to be used in commercial products.。

In addition to Meta, other AI companies, such as French startup Mistral, are also studying the possibility of releasing open-source versions of their technology.。OpenAI, which previously released open-source AI models for voice and image recognition, said its team is considering developing open-source LLMs, but only if they can minimize the risk of abuse.。

"We have a choice between thinking AI is too dangerous a technology to stay open and locking it up and putting it in the hands of a few companies that can control it," Yang said.。"