Overview

Brett Adcock of Figure demonstrates how neural networks have completely replaced traditional robotics programming, with their humanoid robots now running entirely on end-to-end AI. The interview showcases Figure’s manufacturing facility where hundreds of robots operate autonomously, representing a major leap toward general-purpose robotics that could transform the global economy.

Watch the Video

Key Takeaways

  • Once one robot learns a task, every robot in the fleet instantly knows it - unlike humans who must learn individually, creating exponential knowledge scaling
  • Neural networks enable completely different behavior emergence you could never code - robots using hips to close dishwashers and feet to operate appliances naturally
  • Most robotics demonstrations are teleoperated or open-loop programming, not true AI - real autonomy requires uncut footage of closed-loop neural network control
  • Data accumulation creates an insurmountable competitive moat - the more diverse experiences robots have, the better they generalize to new tasks
  • General-purpose robotics represents a $50 trillion market opportunity that will create abundance rather than just replace jobs

Topics Covered