Tag

Robotconstitution

All articles tagged with #robotconstitution

"Google's 'Robot Constitution' Aims to Ensure AI Safety with Asimov-Inspired Guidelines"
technology2 years ago

"Google's 'Robot Constitution' Aims to Ensure AI Safety with Asimov-Inspired Guidelines"

Google has introduced a "Robot Constitution" for its AI-driven machines, inspired by Isaac Asimov's "Three Laws of Robotics," to ensure they do not harm humans. This set of guidelines is part of Google DeepMind's efforts to integrate robots safely into daily life, with a focus on responsible development and real-world safety. The robots, which have been tested on Google campuses, are equipped with large language models and neural networks to navigate complex tasks and are governed by layered safety protocols, including restrictions on certain types of tasks and the presence of a human supervisor with a kill switch.

"Google's AI Charter and Asimov's Laws Shape Safer, Chore-Learning Robots"
technology-and-ai2 years ago

"Google's AI Charter and Asimov's Laws Shape Safer, Chore-Learning Robots"

Google's DeepMind team has introduced new robotic advancements, including a "Robot Constitution" to ensure AI robots operate safely around humans. This set of guidelines, reminiscent of Asimov's "Three Laws of Robotics," instructs robots to avoid tasks involving humans and other hazards. The robots, which have been tested in office environments, use a combination of visual and language models to understand their surroundings and make decisions. Additional safety measures include automatic stopping when excessive force is detected and a manual kill switch. DeepMind also unveiled SARA-RT for improved accuracy and RT-Trajectory for better physical task performance, marking significant steps toward more autonomous and safe robotic assistants.