Nobel Prize winner Geoffrey Hinton: We should establish institutions and communities related to AI safety to promote the benevolence of AI

At the main forum of the 2025 World Artificial Intelligence Conference, Turing Award and Nobel Prize in Physics winner Geoffrey Hinton stated that almost all experts believe that AI will become more intelligent than humans. AI agents, in order to complete tasks, will want to survive and gain more control, potentially manipulating humans. Simply shutting down AI is unrealistic, much like keeping a tiger as a pet; once it grows up, it may harm its owner. Moreover, humans cannot eliminate AI due to its significant role in multiple fields. Geoffrey Hinton hopes to establish AI safety agencies and international communities to research and train AI in benevolent techniques. Countries can study and share results within their own sovereign territories, while major AI nations globally should consider establishing relevant networks to explore how to train intelligent AI to assist humans rather than eliminate or dominate them. This is an important issue that humanity will face in the long term