人工智能医学诊断系统的道德责任研究——以责任空场问题为中心

Research on Moral Responsibility of Artificial Intelligence Medical Diagnostic System:Focus on the Problem of Responsibility Vacancy

医学人工智能的应用给道德责任的划分带来了巨大的困难,其中的一个严峻问题,就是责任空场问题,即每个人似乎都没有过错,因而没有人会承担责任。通过尝试区分直接责任与角色责任,并引入弗洛里迪的分布式责任,可以解决道德责任空场问题。直接责任是错误作为所导致的责任,而角色责任是失职不作为导致的责任。不作为可能导致损害,也可能不导致损害,这属于道德运气的范畴,但如果导致损害,一个人需要为失职的不作为承担全部责任。如果既不存在直接责任,也不存在角色责任,所有的坏运气就由分布式责任兜底。角色责任和分布式责任的提出可以解决责任空场问题。医学人工智能的设计者、数据标注者、监管者和使用者,如果因为不作为导致了不幸的人工智能医疗事故,那么就需要承担角色责任,承担坏运气的后果,如果是一种无人负责的不幸,就需要一种分布式责任来共同承担坏运气的后果。

The application of medical artificial intelligence has brought great difficulties to the division of moral responsibility.One of the serious problems is the problem of responsibility vacancy,that is,everyone seems to be without fault,so no one will take responsibility.This study attempted to distinguish direct responsibility and role responsibility,and introduced Floridi’s distributed responsibility to solve the problem of moral responsibility vacancy.Direct responsibility is the responsibility caused by wrong actions,while role responsibility is the responsibility caused by negligence and inaction.Inaction maybe lead to damage,or not,which belongs to the category of moral luck,but if it does,a person needs to bear full responsibility for negligence.If there is neither direct responsibility nor role responsibility,distributed responsibility pays for all the bad luck.The presentation of role responsibility and distributed responsibility can solve the problem of responsibility vacancy.For designers,data annotators,regulators,and users of medical artificial intelligence,if their inaction leads to an unfortunate medical accident of artificial intelligence,they should assume the role responsibility and bear the consequences of bad luck,while if no one is responsible for the misfortune,the distributed responsibility is needed to bear the consequences of bad luck.