生成进路下人工智能的道德主体地位

Artificial Intelligence’s Moral Agency Under the Enactive Approach

人工智能道德主体地位的确立会撼动现代社会法律及伦理秩序。学术界对人工智能道德主体地位以意向为轴的多元主张仍存在内在张力。生成进路通过“道德知觉”、有机图式分别对道德主体中的道德来源、道德目的进行建构,以强调心智与生命在道德主体形塑过程中的不可或缺性,确立了能感入他者道德世界的公共世界之非自我中心物及时间性的存在者才有资格成为道德主体。基于认知科学的生成进路道德观继承了该进路试图在心智与生命间的解释鸿沟上取得进展的传统,与离身的精神及无意的行为进行二分,提出了整合心智与生命的“道德知觉-有机图式”的道德观,这是一种与古典进路和具身进路相区别的生成进路的人工智能道德主体考量标准。

The establishment of artificial intelligence’s moral agency will greatly influence the laws and ethics of modern society. There are still inherent tensions in the academic community, regarding the multiple propositions on AI’s moral agency where the intent plays a core part. The enactive approach constructs the agent’s source and intent of morality based on “moral perception” and organic schemas, which emphasizes the indispensability of the mind and the lived body in shaping moral agents. Going further, it confirms that only those beings rooted in time and non-egocentric in the common world, who can perceive the moral world of the other, are qualified to be moral agents. Based on cognitive science, this enactive approach to morality inherits its tradition of trying to bridge the explanatory gap between mind and life, and separates the disembodied spirit and unintentional behavior. More importantly, it puts forward the standard of “moral perception-organic schemas”, which is different from the classical approach and embodied approach, to address the consideration of AI’s moral agents.