UC Berkeley Roboticist Warns '100,000-Year Data Gap' Is Keeping Humanoid Robots From Matching AI's Rapid Rise

A UC Berkeley roboticist warns that humanoid robots face a massive '100,000-year data gap' preventing them from matching AI's rapid progress, pushing back against tech leaders' bold claims that robots will soon transform industries — while basic tasks like picking up a wine glass remain unsolved challenges.

• UC Berkeley roboticist Ken Goldberg warns that a '100,000-year data gap' is preventing humanoid robots from gaining real-world dexterity at the same pace AI chatbots have mastered language, pushing back against claims from tech leaders like Elon Musk that humanoid robots will transform industries within a few years.
• Robots continue to struggle with basic physical dexterity tasks such as picking up a wine glass or changing a light bulb, a challenge known as Moravec's paradox, and current methods like teleoperation and simulation have failed to generate enough quality data to solve the problem.
• Robotics is experiencing a fierce paradigm shift between traditional engineering-based approaches and a new data-driven wave, with Goldberg arguing that blue-collar trade jobs remain safe from automation for the foreseeable future while certain white-collar and administrative roles face nearer-term disruption from AI.