Eye On A.I. #303 Fei-Fei Li: Spatial Intelligence, World Models & the Future of AI
32 snips
Nov 23, 2025 
Fei-Fei Li, a pioneer in computer vision and AI, discusses the groundbreaking concept of spatial intelligence and its potential to reshape AI's interaction with the world. She explains how her project, Marble, creates persistent 3D spaces that integrate multiple inputs, not just text. Li delves into continuous learning, the challenges of long-term memory in AI, and why understanding AI differs fundamentally from human perception. She envisions a future where AI combines perception, planning, and imagination, transforming fields from robotics to creative industries.
AI Snips
Chapters
Transcript
Episode notes
Spatial Intelligence Is The Next Frontier
- Spatial intelligence connects perception, action, and reasoning beyond static images and text.
- Fei-Fei Li argues this multimodal understanding is essential for embodied and ambient AI.
Combine Implicit And Explicit Representations
- World models require both implicit and explicit representations to be broadly useful.
- Marble outputs explicit 3D while retaining implicit internal representations for versatility.
Train With Multimodal Inputs
- Use multimodal inputs (text, images, video, coarse 3D) when training world models.
- Marble already accepts varied inputs to deepen spatial learning and real-world usefulness.