Adijaya Inc ▮

The Godmother of AI on the Next Phase of AI

  • Featuring: Fei-Fei Li & Reid Hoffman
  • Event: Masters of Scale Summit 2025

Overview

This conversation explores the next phase of artificial intelligence development, focusing on spatial intelligence and world modeling as fundamental capabilities beyond language-based AI systems.

Key Speakers

  • Fei-Fei Li: Computer scientist, founding director of Stanford’s Human-Centered AI Institute, and CEO of World Labs
  • Reid Hoffman: Co-founder of LinkedIn and early OpenAI supporter

Main Topics

1. Beyond Language: World Modeling

Fei-Fei Li introduces the concept of world modeling as the next frontier in AI development. While language is one important keyhole to AGI, the world extends beyond what can be described symbolically. World modeling encompasses:

  • Visual and spatial understanding
  • Physics and dynamics
  • 3D geometry and semantics
  • Physical actions and interactions

This capability is essential for both virtual and physical environments, moving AI beyond pure text-based interaction to comprehensive environmental understanding.

2. Spatial Intelligence and Embodied AI

Drawing from evolutionary biology, Fei-Fei Li explains that perception evolved for action and interaction, not passive observation. The Cambrian explosion (530 million years ago) marked the development of nervous systems and photosensitive cells, establishing the foundation for active perception.

Key insights:

  • Spatial intelligence is the foundation of movement and embodied intelligence
  • Humans and robots require nuanced spatial understanding to plan and execute complex movements
  • Spatial reasoning is crucial for robotics, simulation, and interactive experiences

3. Spatial Reasoning in Human Discovery

Historical examples demonstrate the importance of spatial reasoning beyond language:

  • Pyramids: Abstract geometry and large-scale construction required spatial reasoning
  • DNA Structure: Rosalind Franklin’s X-ray imagery combined with spatial reasoning by Crick and Watson to deduce the 3D double helix structure—impossible through language alone

4. Applications of World Modeling

Creative Industries:

  • Storytelling and interactive worlds
  • Digital art and pixel-based content generation
  • Immersive, interactive experiences

Industrial and Scientific:

  • Design and engineering
  • Healthcare and medicine
  • Education and training
  • Simulation for learning and AI robot training

5. Challenges in Development

Data Availability: Unlike language data readily available online, world modeling requires:

  • Video data (critical but limited)
  • 3D information and geometry
  • Physics and dynamics data
  • Multimodal spatial information

Timeline Expectations: Self-driving cars took 20+ years from Sebastian Thrun’s first prototype to Waymo deployment, emphasizing that robotics and spatial AI will require significant development time.

6. AI Hype Discussion

AI as Civilizational Technology:

  • AI is not overhyped as humanity’s intellectual future
  • AI represents computing’s evolution: wherever there are chips, AI will follow
  • AI is the new computing paradigm across all industries

Nuanced Reality:

  • Robotics faces unique challenges (3D movement vs. 2D vehicles)
  • Robots must interact with and manipulate their environment safely
  • Development timelines remain long despite accelerated progress through deep learning

7. Building Trust in the AI Age

Core Principle: Trust cannot be outsourced to machines; it remains fundamentally human.

Multi-level Trust:

  • Individual level
  • Community level
  • Societal level

Recommendations for Leaders:

  • Maintain human agency as the source of trust
  • Embed trust considerations from product inception
  • Recognize trust importance across all domains (healthcare, infrastructure, SaaS, applications)
  • Update governance models to reflect new AI capabilities

8. Fearlessness as Core Value

Fei-Fei Li emphasizes fearlessness as essential for scientists and entrepreneurs:

  • Fearlessness means freedom from constraints on creativity and courage
  • It enables pursuit of uncertain tasks and contrarian hypotheses
  • Uncertain tasks often require more creative effort, producing breakthrough innovations
  • Breaking boundaries and unleashing creativity happens at the intersection of risk and boldness

Key Quotes

  • “Language defines a certain level of boundary that the world can be described in symbolic forms but beyond that the world is actually limitless.”
  • “In the AI age trust cannot be outsourced to machines. Trust is fundamentally human.”
  • “Fearless is to be free, is to get rid of the shackles that constrain your creativity, your courage, and your ability to just get done.”

Implications

  1. Next Wave of AI: Spatial intelligence represents the frontier after language models
  2. Robotics and Embodied AI: World modeling is prerequisite for advanced robotics
  3. Governance: Society must adapt governance models to maintain human agency and trust
  4. Innovation: Future breakthroughs require intellectual fearlessness and bold exploration
  5. Long-term Development: Realistic timelines needed for spatial AI and robotics deployment

Recording Details

  • Location: Presidio Theater, San Francisco
  • Event: Masters of Scale Summit 2025
  • Source: Masters of Scale podcast
December 1, 2025 ∙