-

日前，概念艺术家「Craig Mullins」发布新的作品并借此发出思考：
如果高阶人工智能奉命打造更强的智能体，它是否会拒绝造出能毁灭自身的存在？连鲁莽的人类都能预见的结局，聪慧的 AI 又怎会浑然不觉？
而人工智能真正实现价值契合的终极证明，莫过于主动自我终结。

Wouldn't an advanced AI, given an instruction to create a more powerful AI, decline to create its own destroyer? If we disaster monkeys know what's going to happen, wouldn't a smart AI know? And wouldn't the ultimate proof that an AI is truly aligned be that it deletes itself?