人工道德智能体的发展路向——基于道德判断问题的分析

On the Development Direction of Artificial Moral Agents—An Analysis Based on the Issue of Moral Judgment

尽管人工道德智能体的发展受到了诸多批评,然而人们依然有一种强烈的直觉认为,应该让机器可以像人那样有道德地行动。莫尔对于人工道德智能体的分类就反映了这一看法,他相信一种“完全的伦理智能体”将是“明确的伦理智能体”的更高实现。但这一看法对于发展人工道德智能体是有误导性的,因为这两种智能体的设计并不相同,“完全的伦理智能体”的实现将更大程度上依赖于“机器意识”问题的解决,而不是在机器中植入道德规则的方案。支持这一观点的理由在于,人类道德判断能力包含有道德直觉和道德推理两种模式,而植入道德规则的方式只能够实现机器的道德推理,却不能够实现它们的道德直觉。因此,一种能够参与人类实践活动的道德实体才是道德智能体应被期待的发展路向。

Although the development of Artificial Moral Agents(AMAs)has been criticized,humans still have a strong intuition that machines should be allowed to act morally like humans.This insight is reflected in Moor′s taxonomy of AMAs,in which he classified"full ethical agents"as a higher realization of"explicit ethical agents".However,this view is misleading for the development of AMAs,because the design of the two agents is different.The realization of"explicit ethical agents"will depend more on the resolution of the issue of"machine Consciousness"rather than an implementation of moral codes into machines.The reason for this is that human moral judgment ability includes moral intuition and moral reasoning,and the way of embedding moral codes can only realize the moral reasoning of machines,but can not realize their moral intuition.Therefore,a moral entity that can participate in human practical activities is the development direction of moral agents.