Meta is currently working on its own artificial intelligence model, which the company hopes will produce a more powerful AI robot than the OpenAI product GPT-4.

Meta has recently been aggressively gathering AI training chips and building data centers to build a more powerful new chatbot that it hopes will be as high or even higher than OpenAI’s GPT-4. The company reportedly plans to begin training a new large language model in early 2024.

Meta has purchased more Nvidia H100 AI training chips and is upgrading its infrastructure so it no longer has to rely on Microsoft’s Azure cloud platform to train the new chatbot. The company reportedly assembled a group earlier this year to build a model aimed at speeding up the creation of AI tools that mimic human expressions.

Meta faces stiff competition in the generative AI space. OpenAI has stopped training GPT-5 for the moment and won’t for a while, but Apple is said to be pouring millions of dollars every day into its own AI model, Ajax. Google and Microsoft have both expanded the use of artificial intelligence in their productivity tools and Google wants to use generative AI in the Assistant. Amazon has generative AI initiatives underway that could find an outlet in the chatbot Alexa.

Source: The Verge