Coinspeaker
Meta Working on More Powerful AI System that Will Surpass OpenAI’s GPT-4
Meta Platforms Inc (NASDAQ: META) is reportedly developing a new open-source AI system that is expected to be more powerful and advanced than the GPT-4 model offered by OpenAI. Besides, its capacity will be twice as high as Llama 2 – the latest AI model released by Meta and Microsoft Corporation (NASDAQ: MSFT) in July.
As reported by the Wall Street Journal, the training of the new system is expected to start as early as 2024. Recently, Meta increased its investment in AI chips made by Nvidia Corporation (NASDAQ: NVDA), which means it is going to train the new model using Nvidia’s H100s chips. The latest are often described as the world’s first “generative” artificial intelligence chips that serve as catalysts for AI technologies.
According to those familiar with the matter, the new AI system developed by Meta will allow companies to develop services that produce sophisticated text, analysis, and other output. Compared to Llama 2 which was launched on Microsoft’s Azure cloud computing platform, the new Meta’s system will run its own infrastructure. The system will be free of charge.
Meta’s Shift to AI Innovation
With the ongoing work on an AI system, Meta is making a strategic move, shifting to AI innovation. Its upcoming system is expected to challenge not only OpenAI but also the broader AI community where the competition is extremely high.
Just a few weeks after OpenAI released ChatGPT, Google and Microsoft jumped onto the generative AI bandwagon with the release of their own AI chatbots – Bard and Bing respectively. Months later, Apple Inc (NASDAQ: AAPL) started working on AI tools, reportedly spending millions of dollars a day on AI development.
For Meta, working on a new AI model is the next step to achieve its AI ambitions, especially at a time when effective generative AI tools are in demand by most businesses and firms.
Among the latest Meta’s AI innovations was SeamlessM4T – an AI-powered model that allows users to communicate effortlessly through speech and text across different languages. Besides, in August, Meta launched Code Llama – a large language model (LLM) for using text prompts to generate and discuss code.
This month, Meta is reportedly planning to release a new range of AI chatbots with distinct personalities. These AI-powered tools, known internally as “personas”, will feature various characters and provide users with a new and exciting search function with customized recommendations.
Meta CEO Mark Zuckerberg did not respond when asked to comment on the news. However, it is known that on September 13, Zuckerberg will attend a tech forum organized by Senator Chuck Shimmer to discuss the development of artificial intelligence and its regulation. The forum will address such issues as the protection of workers, national security and copyright, and defense against “doomsday scenarios”.
next
Meta Working on More Powerful AI System that Will Surpass OpenAI’s GPT-4