IMPORTANT POINTS:
- Meta is developing a new open-source AI model to rival OpenAI.
- The company purchases H100 semiconductor chips from Nvidia and plans to train the model on its infrastructure.
- Large technology companies and governments are investing and competing in the development of advanced AI systems.
Meta (FB), the parent company of social media platforms Facebook and Instagram, is working on developing a new artificial intelligence (AI) model that promises to rival OpenAI’s more advanced model, according to an exclusive report from the Wall Street Journal.
According to the Wall Street Journal, individuals close to the situation claim that Meta is seeking to make this new AI model “several times” more powerful than its Llama 2 model, released earlier this year.
Additionally, sources said Meta has plans for an open-source system that other companies can use to create AI tools that generate high-level text, analysis and other results.
On the other hand, it is important to mention that Llama was trained on 70,000 million parameters. Although OpenAI has not published its parameters for GPT-4, they are estimated to be around 1.5 trillion.
Preparations underway
However, beyond the parameters, the company has also been building the data centers necessary to develop such a high-level system. In addition, they have acquired more H100 semiconductor chips from Nvidia, considered the most powerful and coveted currently on the market.
The same sources noted that training for this large-scale natural language model (LLM) is expected to begin in early 2024 and be ready for launch sometime next year. It is likely to be released after Google’s Gemini LLM.
It is crucial to remember that Microsoft, OpenAI’s main backer, also collaborated with Meta to make Llama 2 available on Azure, its cloud computing platform. However, Meta is said to plan to train its next model on its own infrastructure.
This news comes at a time when large technology companies and governments are competing to create, deploy and control high-level AI systems.
Global advances in AI
Recently, the UK government announced that it plans to invest $130 million in high-powered chips to develop AI systems.
On another continent, China recently implemented its new AI legislation. Since then, the CEO of Baidu, a major China-based technology company, has stated that more than 70 AI models have been launched in the country.
