Meta is building an AI model to rival OpenAI’s most powerful system


Meta, the guardian firm of social media platforms Facebook and Instagram, says it’s growing a brand new synthetic intelligence (AI) model that can rival the most superior model from OpenAI, in accordance to a Wall Street Journal exclusive.

The Wall Street Journal reported that people aware of the matter stated Meta goals for the brand new AI model to be “several times” extra powerful than its Llama 2 model, which it launched earlier this 12 months.

For the second, The Wall Street Journal’s sources stated that Meta’s plans for the brand new system are for it to be open-source and, subsequently, enable different corporations to construct AI instruments to produce high-level textual content, evaluation and different sorts of output.

The firm has additionally been building the information facilities vital to create such a high-level system whereas buying extra of Nvidia’s H100 semiconductor chips, the most powerful and coveted chips presently out there in the marketplace.

Llama was skilled on 70 billion parameters, and whereas OpenAI hasn’t launched its parameters for GPT-4, they’re estimated to be round 1.5 trillion.

Related: Nvidia drops new AI chip expected to cut development costs

The sources stated Meta anticipates coaching for the massive language model (LLM) to start in early 2024 and to be prepared for launch someday subsequent 12 months. It is seemingly to be launched after Google’s forthcoming LLM Gemini.

Microsoft is a major backer of OpenAI and has additionally collaborated with Meta to assist make Llama 2 out there on Azure, its cloud computing platform. However, the sources stated Meta plans to prepare its upcoming model by itself infrastructure.

This improvement comes as main tech corporations and governments are racing to create, deploy and control high-level AI systems.

Recently, the United Kingdom authorities introduced that it deliberate to spend $130 million on high-powered chips to create AI methods.

Across the globe, in China, the nation’s new laws on AI just lately went into impact. Since then, the CEO of Baidu, a significant China-based tech firm, has stated that over 70 AI models have been released within the nation.

Magazine: AI Eye: Get better results being nice to ChatGPT, AI fake child porn debate, Amazon’s AI reviews