The Elimination of the Moral Responsibility Gap:Research on the Moral Responsibility of Algorithmic Media

当下学界关于算法责任的研究存在两种相对立观点———悲观主义与乐观主义。前者认为,在人与智能媒体共同构成的传媒生态中,责任主体不明,追责困难,存在难以弥合的“责任鸿沟”;后者则提出,建构合适的问责方式可以消解鸿沟。这两种观点均预设了责任鸿沟的存在。然而,通过分析道德责任的四种基本构成与责任鸿沟的主要类型发现,责任鸿沟本质上是一种应该归责的“应然”与实际归责的“实然”之间的“错位”。消除该种错位,可以从三个方面入手:设计层面,建构有意义的人类控制体系,“经纬交错”绘制责任地图;应用层面,采用分布式责任与结构式责任,“总分结合”事后追责;伦理教育层面,实施宏观伦理与微观伦理混合课程,“高低兼容”事前预防。三者共同作用,或能够实现智能媒体的负责任使用。 There are two opposing views on algorithmic responsibility, that is, pessimism and optimism. The former holds that there is an unbridgeable “responsibility gap” in the media ecology composed of human beings and algorithmic media because of difficulty in attributing moral responsibility to agents, while the latter proposes that the gap can be bridged by adopting appropriate methods. Both of these views assume the existence of the responsibility gap. We discuss the four basic components of moral responsibility, deconstruct the three interconnected main types of the responsibility gap, and point out that moral responsibility can only be regarded as a kind of “normative mismatch” between what it should be and what it is in reality in terms of responsibility attribution. To eliminate such mismatch and promote the responsible use of algorithmic media, we can construct the scientific control system, pursue the combination of distributed responsibility and structural responsibility, and implement the mixed courses of macro-ethics and micro-ethics from the aspects of design, application and ethical education respectively.