Ken Goldberg (roboticist) | Armchair Expert with Dax Shepard

Robotics and AI development often suffer from over-optimism, as the field struggles to replicate human physical dexterity despite rapid progress in computational intelligence. Ken Goldberg, a distinguished roboticist and artist, explains that while machines excel at high-level tasks like chess, they remain remarkably clumsy at basic physical interactions, such as folding laundry or clearing a table. This disparity, known as Moravec’s paradox, highlights the immense complexity of human motor evolution compared to recent advancements in data-driven AI. Goldberg argues that current robotics hype often ignores the reality of high failure rates in laboratory settings, emphasizing that true progress requires a complementary approach where machines handle repetitive tasks while humans manage nuanced, unpredictable environments. His work, including the development of DexNet, demonstrates that while deep learning has improved grasping capabilities, achieving human-level physical intuition remains a significant, unsolved challenge.