Fei-Fei Li’s new spatial intelligence startup, World Labs, recently emerged from stealth mode with $230 million in funding. Even without a product, the San Francisco-based outfit is valued at a billion.
The strong support from well-known investors shows their confidence in World Labs and Li’s vision for artificial intelligence that interacts with three-dimensional worlds.
Li is also known as the Godmother of AI.
Born in Beijing in 1976, Li immigrated to the United States as a teenager.
She studied physics at Princeton, electrical engineering at Caltech, and became a leading researcher at Stanford University, which is highly regarded for its AI contributions. That’s where she co-created ImageNet, a massive collection of more than 14 million labeled pictures that became a crucial tool for training AI systems.
The annual ImageNet Challenge, started in 2010, became one of the most important events for artificial intelligence and machine learning. The Large-Scale Visual Recognition Challenge inspired teams from around the world to build the best AI models for recognizing and classifying images. The 2012 challenge showed that deep learning methods could greatly improve model accuracy, facilitating the deep learning revolution.
Before ImageNet, AI struggled to understand objects in images. The tool solved this by labeling its wide range of images with descriptions so computers could learn to recognize them. The dataset fueled many advanced applications like object detection, facial recognition, and autonomous vehicles.
She founded the Stanford Human-Centered AI Institute in 2019 to champion ethical AI.
Today, at World Labs, she aims to create AI that generates immersive virtual worlds with realistic physics and detail. World Labs believes AI can help humans build better worlds through spatial intelligence to create and understand three-dimensional spaces.
They’re developing so-called Large World Models that move beyond current AI models that only handle 2D images and text. They want to make AI tools that allow users like artists, designers, and engineers to create and edit 3D worlds with realistic physics and details.
To do this, AI needs a lot of data from various sources like 3D scans of environments, videos, and images from the real world. It would also use data about how objects behave and interact from simulations and virtual environments.
The potential use cases are vast. Creatives could develop highly detailed 3D environments for movies, games, virtual reality, and augmented reality experiences. Engineers and urban planners could use it to model and simulate buildings or cities to explore layouts virtually before construction begins.
For robotics and autonomous systems, the tech could help machines better understand and navigate the real world around them for improved safety and efficiency.