Fei-Fei Li's World Labs Launches with $230M Funding, Valued at $1 Billion
World Labs, an innovative AI startup founded by Fei-Fei Li, has officially launched, boasting a valuation exceeding $1 billion and securing over $230 million in funding.
Fei-Fei Li, known as the “Godmother of AI” for her pioneering work in generative artificial intelligence and her role as co-director of the Stanford Institute for Human-Centered Artificial Intelligence, co-founded World Labs. The company is focused on advancing "spatial intelligence" systems.
Revolutionizing AI with Spatial Intelligence
World Labs introduces its core technology, known as “Large World Models” (LWMs). In its launch announcement, the company emphasized that current generative AI models are limited to interacting through text, audio, and video. In contrast, humans perceive the world in three dimensions and experience interactions over time.
“To push beyond the limitations of existing models,” the World Labs team stated, “we require spatially intelligent AI that can understand and reason about objects, places, and interactions in a 3D environment.”
The company's mission is to develop AI capable of creating and manipulating virtual spaces with a comprehensive understanding of physics and semantics, effectively bridging the gap between two-dimensional AI models and those that perceive and interact with three-dimensional worlds.
Transforming AI Capabilities
World Labs’ vision could be seen as a “reverse metaverse” for AI. While traditional metaverse technologies are often employed to simulate hazardous environments for human training, World Labs aims to create AI models that can explore and interact with virtual or real 3D environments, potentially revolutionizing fields such as robotics and virtual reality.
The ultimate goal appears to be developing an AI model with the capability to generate entire 3D virtual worlds, akin to how large language models like ChatGPT produce text. This advancement could significantly enhance the functionality and intelligence of AI systems across various domains.
