有限整合论:人工智能的道德决策问题新解  

Limited Aggregation:A New Solution to the Artificial Intelligence Problem of Moral Decision-making

作　　者：孙岩[1] 房帅帅 SUN Yan;FANG Shuai-shuai(School of Philosophy,Shanxi University,Taiyuan 030006,China)

机构地区：[1]山西大学哲学学院,太原030006

出　　处：《自然辩证法研究》2024年第9期83-90,129,共9页Studies in Dialectics of Nature

基　　金：国家社会科学基金一般项目“当代西方正义理论的新发展动态与批判研究”(22BZX014)。

摘　　要：人工智能的快速发展为一种能够适用于不确定性条件的道德决策理论提出了迫切需求。如何合理地将行动者有关的道德因素整合到道德决策过程当中,是当前算法伦理亟需解决的重要问题。不同于整合论以及非整合论两种主要理论,有限整合论通过将道德相关作为对整合策略的限制,实现了对人们具体的行为动机和道德理由更为准确的解释。明确在非道德相关的情形中,数量以及整合策略并不是决定行为是否应当的关键要素。通过特殊的契约主义建构,有限整合论可以为不确定性条件下的道德决策问题提供一种值得思考的道德应当基础。The rapid development of artificial intelligence presents an urgent need for a moral decision theory under uncertain conditions.How reasonably aggregate different moral factors into the moral decision-making has become an important problem that the algorithm ethics needs to solve.Different from aggregation and non-aggregation,by taking moral relative factors as the limitation of aggregation,limited aggregation provides a more accurate explanation of agent’s specific behavioral motivations and moral reasons.Clearing that in non-relative moral situations,quantity and aggregation are not the key factors determining whether an action is moral right.Through a special contractarianism construction,limited aggregation can provide a considerable moral basis for moral decision-making.

关 键 词：不确定性 道德决策 有限整合 

分 类 号：N031[自然科学总论—科学技术哲学]