Waymo Co-CEO Dmitri Dolgov: The Demo Is Only 1% Of The Work

Y Combinator

Summary

Dmitri Dolgov, Waymo Co-CEO, discusses the seven key lessons learned over the years in building and safely deploying AI in the physical world, focusing on autonomous vehicles. He highlights the significant differences and challenges of physical AI compared to digital AI, emphasizing the need for robust, safety-first systems. The presentation covers the long journey from a working demo to a scalable product, the importance of choosing the right technology, continuously adapting to new innovations, leveraging structured end-to-end models, and the critical role of high-fidelity simulation. Dolgov concludes by outlining Waymo's AI ecosystem, the flywheel effect of data and evaluation, and the strategic importance of metrics and earned trust for real-world physical AI deployments.

Chapters

  1. 0:07Introduction to Physical AI and Waymo
  2. 0:23Physical AI vs. Digital AI
  3. 2:23Waymo Driver's Impact
  4. 2:32Waymo's Scale and Safety Record
  5. 3:23The 'Move Fast and Ship Safely' Mantra
  6. 3:46Massive Opportunity in Physical AI
  7. 3:51Four Gaps of Physical AI
  8. 6:27Defining Operating Conditions
  9. 8:08Lesson 1: The Gap Between Demo and Product
  10. 8:21Demo is 1% of the Work
  11. 10:39Waymo's Journey: Demo to Product
  12. 11:44Exponential Ladder of Nines
  13. 12:21The Long Tail Problem
  14. 14:01Hype Cycles and Real Products
  15. 14:43Count Your Nines, Not Demo Views
  16. 14:48Lesson 2: Choosing the Right Technology
  17. 14:53Architecture Dictates Nines
  18. 15:01Performance vs. Effort Curves
  19. 16:52Multi-Modal Sensing for Superhuman Performance
  20. 19:25Redundancy in Sensing
  21. 20:19Hardware Evolution and Cost
  22. 21:05Lesson 3: Riding Tech Waves with Unification
  23. 21:13Continuously Ride Tech Waves
  24. 23:00From Research to Production Without Regression
  25. 24:44Waymo Foundation Model
  26. 24:55Multimodal World Action Language Model
  27. 26:35System One (Fast) and System Two (Slow) Paths
  28. 29:16Benefits of a Foundation Model
  29. 29:57Lesson 4: The 'Bitter Lesson' and Structure
  30. 30:06General Methods Beat Handcrafted Knowledge

Topics covered

  • Challenges of Physical AI
  • Demo vs. Product Reality

+8 more topics (full content when you follow the channel)

Transcript excerpt

Good afternoon, everyone. It's great to be here. Uh we talk a lot about AI that lives on your screen, uh lives in the digital world. And today, I'd like to talk to you about a different kind of AI that we've been building at Waymo. AI that lives in the real physical world. How many of you, by the way, have been in a Waymo? Just raise your arms. Wow, okay. That is impressive. Especially, I understand many of you are out of town. Uh the folks who are visiting and have not had a chance to check out Waymo, I hope you while you're here in the Bay Area, give it a try. Uh so, this being a startup school, I structured this presentation as a sequence of lessons. Seven lessons that we've learned over the years at Waymo around what it takes to build and safely ship today's most mature application of AI in the physical world, the Waymo driver. Uh let me start with a short video. Uh this is a clip fr…