How to be 'fearless' in the AI age, with Fei-Fei Li and Reid Hoffman

What It Covers

Fei-Fei Li explains why spatial intelligence represents the next frontier of AI beyond language models, discussing her company World Labs' work on world modeling and its critical applications in robotics, simulation, and creative industries.

Key Questions Answered

• •Spatial Intelligence Foundation: World modeling extends beyond language to represent visual semantics, physical space, and actions—enabling immersive experiences for creators, designers, and industrial applications including healthcare, education, and robotic simulation that language alone cannot express.
• •Data Challenge in World Modeling: Unlike language data readily available on the internet, world modeling requires multimodal spatial data including three-dimensional geometry, physics, and dynamics that are significantly harder to obtain, creating a fundamental bottleneck for development progress.
• •Robotics Timeline Reality: Self-driving cars took over twenty years from Sebastian Thrun's 130-mile Nevada desert demonstration to Waymo operating in San Francisco—even with established automotive infrastructure. More complex robots that must touch objects precisely will require substantially longer development cycles.
• •Trust as Human Responsibility: Trust in AI cannot be outsourced to machines—it must remain fundamentally human at individual, community, and societal levels. Entrepreneurs should prioritize human agency and trust-building from day one, regardless of whether their product seems directly connected to sensitive applications.