机器伦理何以可能:现有方案及其改良

How can Machine Ethics Be Possible:Existing Schemes and Improvements

人工智能要成为真正意义上的智能,一个不可回避的问题就是如何使智能机器的行为更符合人类意义上的伦理道德,这也是作为人工智能新兴研究领域——机器伦理的核心内容。功利主义和义务论都将道德推理统一于一个总原则之中,对机器伦理研究有着重要的启发意义。功利主义预设了一种拉普拉斯式的万能神灵理性观,在应用于机器伦理时,需要通过"有限理性"与"捷思法"对之进行改良。义务论对伦理中"应然"维度和规范性的强调,使机器难以有效解决道德规则冲突,因此,义务论进路在整体上并不适用于当前以计算和遵守规则为核心的智能机器。

If artificial intelligence would be a real sense of intelligence,an unavoidable question is how to make the behavior of the intelligent machine conforms to the ethics of human. Both utilitarianism and deontology put moral reasoning on the basis of general principles,which will have important inspirations on machine ethics. On the one hand,utilitarianism pursues the greatest happiness of the most people. When utilitarianism is applied to machine ethics,we need the ＂ bounded rationality＂ and ＂ heuristics＂ to improve it. On the other hand,deontology is unable to solve the conflicts of moral rules,therefore,deontology approach is not very suitable for current intelligent machines based on calculation and following rules.