We're sunsetting PodQuest on 2025-07-28. Thank you for your support!
Export Podcast Subscriptions
cover of episode Fei-Fei Li: Spatial Intelligence is the Next Frontier in AI

Fei-Fei Li: Spatial Intelligence is the Next Frontier in AI

2025/7/1
logo of podcast Y Combinator Startup Podcast

Y Combinator Startup Podcast
AI Deep Dive AI Chapters Transcript
People
F
Fei-Fei Li
Topics
Fei-Fei Li: 我认为通用人工智能（AGI）离不开空间智能，我想要解决这个问题。我热爱创业，享受从零开始的感觉，专注于构建，不受过去成就或他人看法的束缚。我的职业生涯始于人工智能寒冬的末期，并参与和推动了人工智能的崛起。计算机视觉的发展从物体识别到场景理解，现在扩展到对世界的建模，因此我决定离开学术界，创立World Labs，致力于更具挑战性的空间智能研究。我经常从进化和脑科学中寻找灵感，来确定下一个需要解决的核心问题。解决空间智能问题，理解、生成和推理3D世界，是在3D世界中行动的关键，也是实现通用人工智能（AGI）的根本。这需要创建超越平面像素和语言的世界模型，真正捕捉世界的3D结构和空间智能。我采取混合方法，重视数据量，更重视数据质量，避免垃圾数据。

Deep Dive

Chapters
Dr. Fei-Fei Li discusses the creation of ImageNet, a project that provided the data backbone for modern computer vision. She details the challenges of working with limited data in the early 2000s and the subsequent paradigm shift towards data-driven methods. The ImageNet challenge and the breakthrough moment of AlexNet in 2012 are highlighted.
  • ImageNet conceived almost 18 years before the interview (around 2007)
  • Initial challenges included limited data and non-functional algorithms
  • The paradigm shift to data-driven methods led to the creation of ImageNet
  • The ImageNet challenge helped benchmark machine learning algorithms
  • AlexNet's success in 2012 marked a significant milestone, combining data, GPUs, and neural networks

Shownotes Transcript

A fireside with Dr. Fei-Fei Li on June 16, 2025 at AI Startup School in San Francisco.Dr. Fei-Fei Li is often called the godmother of AI—and for good reason. Before the world had AI as we know it, she was helping build the foundation.In this fireside, she recounts the creation of ImageNet, a project that helped ignite the deep learning revolution by providing the data backbone modern computer vision needed. She walks through the early belief in data-driven methods, the shock of seeing convolutional networks outperform expectations in 2012, and how those breakthroughs led to captioning, storytelling, and ultimately, generative models.Now, she’s taking on one of AI’s hardest frontiers: spatial intelligence. Fei-Fei shares why modeling the 3D world is essential for AGI—and why it may be even more difficult than language.