Shulou(Shulou.com)11/24 Report--

CTOnews.com, June 14 (Xinhua)-- Social media giant Meta (formerly Facebook) announced on Tuesday that it will open some components of a new "humanoid" artificial intelligence model to researchers that can analyze and complete unfinished images more accurately than existing models.

The model, called I-JEPA, uses background knowledge of the world to fill the missing parts of the image, rather than extrapolating only from nearby pixels, as other generative artificial intelligence models do. This method adopts the humanoid reasoning method advocated by Yann LeCun, the chief artificial intelligence scientist of Meta, and helps to avoid common mistakes in image generation by artificial intelligence, such as an extra finger.

Meta CEO Zuckerberg said sharing models developed by Meta with other researchers could help the company promote innovation, identify security vulnerabilities and reduce costs. "for us, if the industry standardizes the basic tools we are using, then we can benefit from improvements from others," he told investors in April. "

CTOnews.com notes that Meta is also starting to incorporate generative artificial intelligence into its consumer products, such as advertising tools that create image backgrounds based on text prompts and Instagram products that modify users' photos.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.