Robotics, Art, and the 100,000-Year Data Gap

“Robotics, Art, and the 100,000-Year Data Gap” is an interview in which Jim Euchner speaks with roboticist-artist Ken Goldberg about why robotics progress looks so different from the recent “ChatGPT moment” in generative AI. Goldberg’s central point is that robotics faces a massive training-data shortage compared to LLMs: we have enormous amounts of text and images for language models, but only a thin slice of high-quality, real-world interaction data for robots—especially for manipulation and dexterity. Because physical work depends on messy realities like friction, touch, and unpredictable environments, Goldberg argues robotics will advance more steadily, application-by-application, rather than through sudden leaps driven purely by scaling data.