Mark Zuckerberg's Meta working on a new, more advanced AI model to rival OpenAI’s GPT-4
In July, Meta, the parent company of Facebook, Instagram, and WhatsApp, publicly released its AI model Llama 2. Built on 65 billion parameters, Meta described its platform as a “smaller foundational model” that was suitable for AI research and testing new approaches as it consumed less computing power. But now, a new report suggests that CEO Mark Zuckerberg is entering the big AI race and intends to compete with the likes of OpenAI. Meta is apparently working on a new AI model that is far superior to Llama 2 and will rival GPT-4, the latest large language model by OpenAI.
According to a report by the Wall Street Journal, Meta wants its new AI model to be ready for a release next year and expects to begin training the AI model as soon as the first three months of 2024. Interestingly, Zuckerberg is pushing to make the AI model open-sourced and consequently free for businesses to create AI tools, just like its predecessor. Read more: 5 things you should know about Meta's ChatGPT-rival, AI platform LLaMA
The new AI model is being built to have all the capabilities of an LLM including generating sophisticated text, analysis, and humanlike expression. As per the report, Meta has begun buying a higher quantity of Nvidia H100 AI training chips, the most advanced chip produced by the company. The reason behind it is believed to be that Meta wants to train the new model on its own infrastructure unlike Llama 2, which was trained on Microsoft's cloud computing platform Azure.
Earlier, it was reported that Meta built a team of AI researchers to build tools that can bring humanlike experience, and now it is said that Zuckerberg has accelerated the workings of the team. This team has already begun building AI tools for the company, as
Read more on tech.hindustantimes.com
