Turing Award winner and reinforcement learning pioneer Richard Sutton has left John Carmack's Keen Technologies to launch Oak Lab, a Toronto-based startup co-founded with Khurram Javed. Sutton critiques current deep learning methods as weak and inefficient, aiming instead to build autonomous AI agents that learn continuously from their environments in real time. Oak Lab's long-term goal is to develop a trillion-parameter agent that plans and learns using only 20 watts of energy.
Oak Lab founding
- ▪Richard Sutton and Khurram Javed co-founded the Canadian-incorporated startup Oak Lab in June 2026
- ▪Oak Lab is based in Toronto, Canada, and focuses on developing autonomous learning AI agents
Continual learning AI technology
- ▪Oak Lab's long-term goal is to build an AI agent with one trillion parameters that learns and plans in real time using 20 watts of energy
- ▪Oak Lab is developing algorithms that learn in real time without storing or replaying data, aiming to use less processing power
Sutton's career moves
- ▪Richard Sutton joined Keen Technologies in September 2023 after Google's DeepMind closed its Edmonton lab
- ▪Richard Sutton left Keen Technologies, an AI startup founded by John Carmack, to launch Oak Lab
Keen Technologies AGI pursuit
- ▪Keen Technologies' financial backers include Shopify Chief Executive Officer Tobi Lütke
- ▪Keen Technologies is a Dallas-based firm trying to build artificial general intelligence with human-matching or -beating capabilities
Reinforcement learning research
- ▪Richard Sutton is a computing science professor at the University of Alberta
- ▪Richard Sutton, a 2024 Turing Award winner, is a pioneer in reinforcement learning, which trains AI through trial and error
Deep learning rethinking approach
- ▪Richard Sutton and Khurram Javed left Keen Technologies to pursue a path that involves rethinking deep learning
- ▪Richard Sutton critiques current deep learning methods as weak and inefficient, arguing they require a thorough reworking rather than minor tweaks
Story comments
Loading comments…