Talking Robots: Rolf Pfeifer - New AI

Embodied artificial intelligence challenges the traditional computational paradigm by asserting that cognition is not an abstract process but is fundamentally constrained and enabled by physical morphology. Intelligence emerges from the interaction between an agent's body and its environment, where sensory-motor loops provide the raw material for concept formation. Even abstract capacities like mathematical reasoning are grounded in physical experience, as evidenced by conceptual metaphors that link bodily sensations to complex logic. While simulations offer valuable insights, building physical robots remains essential for understanding the dynamics of real-world interaction and the limitations of current sensor technology. Future progress in AI requires an interdisciplinary approach, integrating material science, biomechanics, and neuroscience to move beyond simple algorithms toward systems that truly mirror the complexities of biological intelligence.