Rich Sutton Launches Oak Lab to Revolutionize AI Learning

Rich Sutton Launches Oak Lab to Revolutionize AI Learning

Richard Sutton, winner of the 2024 Turing Award and co-founder of modern reinforcement learning, has launched the startup Oak Lab in Toronto with Khurram Javed. Both previously worked at Keen Technologies, John Carmack's AI company. Sutton describes current deep learning methods as "weak and inefficient" and asserts that they "need not just simple tweaks, but fundamentally new ideas and a complete overhaul before they can provide a solid foundation for achieving the more ambitious goals of AI."

His recent statements reveal his intentions. In June, he argued that generative AI is effective for imitation but incapable of evaluating its own results, rendering it unfit for true discovery. He aims to build AI agents that continuously learn from their environment, construct internal models of the world, and autonomously manage variation, evaluation, and selection. Oak Lab is founded on this idea. Like Keen, the company bets on reinforcement learning and the belief that AI must learn from experience during operation rather than being trained once on static datasets. The long-term goal is to create an agent with one trillion parameters that learns and plans in real-time using 20 watts of energy.