Elephants Still Don't Play Chess
Abstract
What might the standard model of artificial intelligence look like? While today’s systems are built from data and language, natural intelligence emerges from brains in bodies exploring the world. In this talk, I argue that true progress in AI will come from placing robots at the centre, and allowing agents to discover their own representations of the world. I will share how this embodied perspective could form the basis of a unifying framework for artificial intelligence, and what it might mean for the future of intelligent machines.
Related to the book chapter Does the Body Matter? Perspectives on the Embodied Nature of Consciousness.
Steven James
Associate Professor
My research focuses on artificial intelligence and reinforcement learning, in particular designing agents capable of learning transferable, high-level abstractions of their environment.