AI 服務暫時不可用，以下為來源正文，待恢復後補全翻譯。
Imagine bringing a robot into your kitchen and saying, “Help me clean up after dinner.” You have just compressed a remarkable amount of engineering into six words. The robot must distinguish leftovers from rubbish, discover where plates belong, and work out why a drawer refuses to close. Eventually, someone will hand it a wineglass. Everyone will suddenly become very interested in the quality of its training data. That kitchen captures the promise of a ChatGPT moment for robotics. We would be able to give a machine useful new work through conversation and examples, with sufficiently little setup that teaching it becomes an ordinary activity. Generalist robot models are making this prospect more credible. The remaining distance involves learning, control, and the economics of getting a machine to work somewhere new. ChatGPT made broad capability easy to explore. You could ask for a poem, then a program, then an explanation, and discover the range yourself. Robotics needs an equally persuasive encounter with versatility. The complication is that its answers have mass. From language to physical action Read more