DeepMind Unveils New Safety Rules and Systems to Make AI-Powered Household Robots More Responsible
- Google DeepMind announced new rules to make AI-powered robots safer when doing chores in homes
- The rules aim to give robots a better understanding of objects and environment using large language models
- DeepMind introduced a "robot constitution" with safety prompts based on Asimov's Three Laws of Robotics
- New systems called AutoRT and SARA-RT are designed to make robots faster at decision-making
- RT-Trajectory system provides robots clearer instructions on physical actions needed to complete tasks