Chelsea Finn: This is the State of the Art in Robotics

Y Combinator Startup Podcast

Aug 13

Overview Shownote Highlights Transcript Chapters Pins

Shownote

Robots can already fold laundry, make espresso, clean kitchens, and assemble things. The harder problem is getting them to do those tasks reliably, for long periods of time, without a human babysitting them. At Startup School 2026, Physical Intelligenc...

Highlights

This conversation explores the technical and practical challenges of building robots that can learn broadly, adapt to unfamiliar situations, and operate independently in the physical world.

06:58

Robots must learn from failure to become reliable.

34:57

General-purpose robots can generalize without post-training.

46:57

Embodied data may limit robotics democratization

53:04

Robots must become faster before they become ubiquitous.

Chapters

How reinforcement learning helps robots recover, remember, and work reliably

00:00

Can diverse data teach one robot to handle many different tasks?

22:56

Why robotics is approaching general-purpose intelligence—and what still holds it back

38:04

The next priorities: faster execution, safer behavior, and dependable real-world performance

49:55

Transcript

Chelsea Finn: Today, I'm going to be talking about the state of the art of Physical Intelligence. And in particular, two years ago, I founded a company called Physical Intelligence. And we're really interested in how we can basically develop any robot, or ...