Key Points

  • Mark Zuckerberg’s Meta is reportedly working on an AI project that is meant to rival OpenAI’s ChatGPT
  • Wall Street Journal Exclusive story indicates that Meta is in very advanced stages of AI chatbot development

Meta has been on the spot before for training its AI with licensed artwork, and reports leaked that it was working on an AI chatbot. A Wall Street Journal exclusive story now indicates that its AI project will rival OpenAI.

Meta’s AI conversation continues.

Meta, the parent company to multiple social media platforms, WhatsApp, Facebook, Instagram, and Threads, has indicated that it is working on the world’s most powerful artificial intelligence project. 

This project is meant to rival OpenAI’s ChatGPT models, with sources close to the company saying it aims to have its AI (Llama) be “several times” more powerful than its earlier version, the Llama 2 model released earlier this year.

Meta has been building data centers necessary for developing such a project, acquiring respectable numbers of Nvidia’s H100 semiconductor chips- the most powerful and highly coveted ones on the market. Wall Street Journal’s sources expect that the company will make the AI project open, allowing other companies to learn from its code and produce similar high-level models.

Llama was trained on over 70 billion parameters, making it one of the most potent AI projects. However, OpenAI’s ChatGPT 4 was reportedly trained on over 1.5 trillion parameters, though official numbers are yet to be released. 

At the same time, the claims of Meta having a rival Llama model remains to be certified as no official communication is out yet. Keep watching Fintech Express for updates on this and other tech-related developments.