Co-director Stanford HAI; Godmother of AI; founder World Labs
NOV 16 2025
The Vision
World Models Are the Next Frontier
"We've given AI language. The next step is giving AI an understanding of the physical world — spatial intelligence at scale."
World Labs' mission: build AI with spatial intelligence, not just language intelligence
ImageNet moment (2012) sparked the deep learning revolution — we're at a new inflection
Spatial AI enables robots, autonomous systems, and physical world automation
The gap: current AI understands text and images but not 3D space and physical causality
Framework
From ImageNet to World Models
2012
ImageNet breakthrough
2017
Transformer era begins
2026+
Spatial AI era starts
ImageNet: AI learns to see (perception)
GPT: AI learns to speak and reason (language)
World models: AI learns to understand and predict physical reality
Each leap required new architecture, new data, new compute scale
Fei-Fei's thesisThe next $10T of AI value comes from the physical world: manufacturing, robotics, medicine, infrastructure.
The Jobs and Society Question
What AI Means for Humans
Fei-Fei's optimism: AI augments human capability — in medicine, education, creativity
Her concern: Transition speed matters more than transition direction — we're moving faster than institutions
On jobs: Physical world AI creates new roles faster than it eliminates current ones
On responsibility: The researchers who build AI must help design the social infrastructure
The HAI mission
Human-Centered AI: technology designed around human flourishing, not just capability maximization.
The responsibility principle
Power comes with responsibility. The AI lab that ignores social impact is borrowing against the future.
Playbook
Think Like an AI Pioneer
Study the physical world problem space — that's where the next generation of AI impact is
Ask: does this AI system increase or decrease human agency? It should always increase it.
Invest in AI education — the shortage of AI-fluent domain experts is the real bottleneck
Support research institutions alongside AI companies — they provide the safety net
The ImageNet lessonOne researcher, one dataset, one breakthrough changed 10 years of AI. Concentrated insight still wins.
Contrarian
AI Pessimism Myths Fei-Fei Addresses
✗AI will make humans obsoleteINSTEAD →✓ AI will make humans more capable. Augmentation has always outpaced displacement in technology history.
✗AI research is too dangerous to continueINSTEAD →✓ AI safety research requires continuing AI development. You can't make safe what you don't understand.
✗The robots are coming for physical jobs firstINSTEAD →✓ Robots are coming for the most dangerous and dull physical jobs first — and that's a good thing.
✗AI is developed only by a few labsINSTEAD →✓ The open-source AI movement is real and growing. Concentration in labs is temporary.