Within 3 Months, 'AI Mother' Fei-Fei Li Creates a Unicorn Startup

On July 17, it was reported that World Labs, a cutting-edge startup founded by renowned Chinese computer scientist Fei-Fei Li, has surpassed a $1 billion valuation. The company focuses on using human-like visual data processing technologies that enable artificial intelligence (AI) to exhibit advanced reasoning capabilities. Since its inception in April of this year, World Labs has completed two rounds of funding, attracting investments from top tech investors including Andreessen Horowitz and Radical Ventures. The latest funding round may reach approximately $100 million.

World Labs represents the latest addition to the list of AI startups receiving significant investment. Data from the reputable analytics firm PitchBook shows that over the past three months, investors have poured more than $27 billion into U.S. AI startups, accounting for nearly half of the total funding for all startups during the same period. Fei-Fei Li, widely regarded as the "Godmother of AI" for her pivotal role in developing the large-scale image dataset ImageNet, has gained prominence in the AI sector. This dataset was instrumental in creating the first generation of computer vision technologies capable of reliably identifying objects.

Li is the inaugural Sequoia Professor in the Department of Computer Science at Stanford University and co-director of the Stanford Institute for Human-Centered Artificial Intelligence (HAI), which she founded in 2019 to improve human welfare through emerging technologies. Li has previously led Google Cloud's AI business and served on the board of social platform X from 2020 to 2022.

At World Labs, the team aims to integrate human-like visual data processing techniques to create "spatial intelligence" within AI. In April, Li delivered a TED talk in Vancouver about the potential for machines to understand and navigate three-dimensional spaces. Her research includes an algorithm that can infer how images and text appear in a 3D environment and act based on those predictions.

To illustrate her concept, Li described a scenario in which a cat pushes a glass cup to the edge of a table. She noted that the human brain can quickly assess the cup's geometric shape, its position in 3D space, and its relationship with the table, the cat, and other objects, allowing us to predict outcomes and prevent mishaps. Li emphasized that "nature has created a virtuous cycle of 'seeing' and 'doing' driven by spatial intelligence."

She envisions a machine capable of understanding the complex interplay of objects in the physical world, facilitated by a vast repository of labeled images that are essential for recent breakthroughs in AI. This knowledge allows for training autonomous vehicles to navigate their environments and helps AI models recognize objects based on visual cues.

The realm of AI innovation is at a crossroads. Some researchers believe that enhancing reasoning abilities can be achieved by building on existing, more complex models, while others advocate for the development of new "world models" that can gather visual information from their physical surroundings to develop logical reasoning—a method akin to how infants learn.

Among the AI startups drawing investor interest, several are developing intelligent robots capable of understanding and manipulating their physical environments. For instance, Skild AI is working on a "universal brain for various robots" and recently achieved a valuation of $1.5 billion following $300 million in funding from notable entities, including SoftBank and Jeff Bezos' investment firm, Lightspeed Venture Partners. Li's focus on "spatial intelligence" aligns with these groundbreaking efforts.