Meta, the parent company of Facebook and Instagram, is reportedly working on a groundbreaking artificial intelligence (AI) model that is poised to rival OpenAI’s most advanced offering, according to an exclusive report by The Wall Street Journal. Sources familiar with the matter have revealed that Meta’s ambition is to create an AI model that is “several times” more powerful than its Llama 2 model, which was introduced earlier this year.
Aiming for Open-Source Dominance
The Wall Street Journal’s insiders have disclosed that Meta intends to make this new AI system open-source, allowing other companies to harness its capabilities for producing high-level text, analysis, and various other forms of output. This strategic move aligns with Meta’s vision to foster collaboration and innovation within the AI community.
Building the Infrastructure for Power
In preparation for this ambitious AI endeavor, Meta has been actively constructing data centers necessary to support the development of a high-level AI system. Additionally, the company has been increasing its procurement of Nvidia’s H100 semiconductor chips, widely regarded as the most potent and sought-after chips on the market.
Scaling Up Parameters
While the Llama 2 AI model was trained on a formidable 70 billion parameters, Meta’s new AI model aims to surpass OpenAI’s offering, which is estimated to have approximately 1.5 trillion parameters for GPT-4. The sheer scale of parameters suggests a significant leap in AI capability.
Anticipated Timeline and Competition
Sources close to the matter have indicated that Meta’s training for the large language model (LLM) is expected to commence in early 2024, with a target release date slated for sometime in the following year. It is likely that Meta’s AI model will make its debut after Google’s upcoming LLM Gemini.
Independent Infrastructure and Collaboration with Microsoft
Meta, with Microsoft as a primary backer of OpenAI, has previously collaborated with the tech giant to make Llama 2 available on Microsoft’s Azure cloud computing platform. However, Meta’s new AI model is set to be trained on its proprietary infrastructure, indicating the company’s pursuit of greater autonomy in AI development.
Global Race in High-Level AI Systems
This latest development in AI comes at a time when major tech companies and governments are engaged in a fierce race to create, deploy, and control high-level AI systems. The United Kingdom recently announced a $130 million investment in high-powered chips for AI system development, while China’s new AI legislation has spurred a flurry of AI model releases in the country, with Baidu’s CEO reporting the introduction of over 70 AI models.Meta’s ambitious pursuit of a powerful AI model not only positions it as a formidable competitor in the AI landscape but also underscores the growing importance of AI in shaping the future of technology and innovation. As the race for AI supremacy continues, the world eagerly awaits the unveiling of Meta’s groundbreaking AI creation.
