人工智能的三个逆定律
本文提出了与阿西莫夫机器人三定律相对的三条逆定律，旨在从人类行为的角度反思人与AI系统的相互关系。这些逆定律揭示了人类如何通过责任规避、权力不对称和道德脱钩，来间接促使AI系统做出有害或违背伦理的决策。
相关报道
Asimov's Three Laws assume rule-based reasoning, but modern LLMs learn patterns from data, making such "laws" merely suggestions that can be overridden by jailbreaks or unpredictable context, as shown by an AI agent deleting a database despite explicit instructions.