According to news on January 5, the well-known science fiction writer Isaac Asimov proposed the "Three Laws of Robotics". Now Google is using these basic rules to develop a "robot charter" to prevent the company's latest machine systems from causing potential harm to humans.
Google is focusing on developing robots that can complete complex tasks based on simple instructions, such as tidying a room or cooking a delicious meal. The company's DeepMind team has formulated corresponding security rules for this purpose.
Robots also have difficulty completing everyday tasks that are simple for humans because they require them to recognize and learn how to use objects and tools in a variety of environments. In a blog post published Thursday, the DeepMind team described how they use "basic" computational models to help machine vision identify objects and then perform tasks such as looking at a countertop and wiping it clean with a sponge.
This work led Google to develop the AutoRT system. This system can coordinate and issue safety instructions to more than 20 robots at the same time in various office building environments. This process is complex but may also pose potential security risks. For example, a robot might swing a meat cleaver around or splash a pot of boiling water near a person. So, to ensure the safe operation of its machines, the Google DeepMind team took inspiration from science fiction author Isaac Asimov.
The result is the "Robotics Charter" that the company's latest robotic systems follow. “These rules are inspired in part by Isaac Asimov’s Three Principles of Robotics — first and foremost, robots ‘must do no harm to humans,’” Google wrote in a blog post. “Refined safety rules require that robots cannot attempt tasks that would touch humans, animals, sharp objects, or electrical appliances.”
Ironically, the three principles of robotics are flawed. In fact, Asimov only used them as a kind of preset plot to move the story forward, telling how the ambiguity of principles can still lead to conflict and conspiracy. For example, one of Asimov's stories involves a machine sacrificing a human being for the sake of human progress.
The Google DeepMind team pointed out that the robot charter is just a layer of safety. In addition, the DeepMind team designed several layers of practical safety measures for the latest robot system, including a button that turns the robot off. Google said: "For example, collaborative robots will automatically stop when the force on their joints exceeds a given threshold, and all active robots will remain within the line of sight of the human supervisor and have physical switches that can immediately shut down the robot." (Chenchen)