a16z· Jul 28, 2026
Fei-Fei Li is Solving the Hardest Problem in Robotics | World Labs with a16z
Fei-Fei Li, cofounder of World Labs, and Yunzhu Li, cofounder of SceniX, discuss spatial intelligence and robotics with a16z host Martin Casado.
No Priors Ep. 117 | With Co-Director of Stanford's HAI & Founder of World Labs Dr. Fei-Fei Li
On No Priors, Stanford HAI co-director and World Labs founder Fei-Fei Li discusses spatial intelligence, robotics, AI research, and human-centered applications with the show’s hosts.
In this talk
- 0:47
Why build spatial intelligence
Li says she started World Labs to build 3D world models that help people understand, generate, and interact with digital environments. She argues spatial intelligence is a core part of human and animal intelligence, and that useful worlds need plausible geometry and physics.
- 9:52
Robotics needs varied forms and data
Li expects people to live alongside robots in many shapes, with form factors suited to different tasks; underwater robots, for example, could be shaped like fish. She sees robot training as a mix of data sources and says simulation is underrated. For manipulation, she emphasizes combining haptics with vision and spatial data.
- 13:23
Creative tools and world model challenges
Li sees 3D creation as a near-term use for spatial AI, helping designers, artists, VFX teams, marketers, and game developers work with AI collaborators. World models could also support interactive XR content and design tasks that lend themselves to reinforcement learning. She identifies limited 3D data and the difficulty of making 3D easy to use as major challenges.
- 19:35
ImageNet and fearless research
Li recounts how her PhD work on object recognition led to a dataset spanning 101 categories, which became ImageNet. She recalls the effort to collect and clean images, including asking her mother to help review them, and credits ImageNet with supporting breakthroughs such as AlexNet. Her advice to researchers and founders is to be bold while staying grounded enough to work through practical constraints.
- 31:58
AI should empower people
Li says human-centered AI should support people, relationships, justice, and prosperity across communities. In healthcare, she points to needs in drug discovery, diagnosis, precision medicine, care delivery, aging, and mental health, arguing that AI can help address gaps in support and scientific discovery.
Briefs summarise talks and interviews; timestamps open the video at that moment. Claims in a talk are the speaker's own.