🤖 Sergey Levine: "Robot Mistakes Are Not Failures, But a Way to Learn"

Sergey Levine is a professor at UC Berkeley and co-founder of Physical Intelligence. His team develops foundational models for robots, enabling them to perform a wide range of tasks—from cleaning the kitchen to managing an entire farm.

In a podcast with blogger Dwarkesh Patel, he shared his predictions about the future of robotics:

➡️ Robots must become universal. Today, individual prototypes can fold laundry or wash dishes. But Levine's goal is to create a general "brain" for robots, one that can adapt to any task.

➡️ The self-improvement flywheel. Once robots are deployed "in the field" and start gathering their own experience, their learning process will accelerate exponentially. Levine refers to this effect as the "flywheel"—a mechanism that gains momentum on its own.

➡️ Robots are becoming more affordable. While research robots used to cost hundreds of thousands of dollars, manipulators can now be purchased for just a few thousand. Mass production also speeds up learning.

➡️ The timeline is closer than it seems. The first robots that are genuinely useful in everyday life could appear within the next few years. Their development path will likely resemble the evolution of code assistants: progressing from simple suggestions to full-fledged projects.

"As an example, a robot might first make you a cup of coffee and later manage an entire café," explains Sergey Levine.

➡️ The power of mistakes. Unlike AI assistants, robots can learn from their physical-world errors. If an android drops a shirt and then picks it up, that's not a failure—it's a new experience gained.

📱 Watch the full interview with Levine here.

#robots #interview @hiaimediaen