AI实体刑事责任主体论的规范证成

Specification in AI Entity Criminal liability Subject

为全面保护法益及预防犯罪,应厘清AI实体是否能够具有刑事责任主体地位这一问题。学界对此形成了肯定论、否定论及终结论,但多数论证呈现出笼统概括、混淆概念及片面否定等不当之处,应从甄别AI实体能否具有自由意志与刑事可罚性展开规范讨论。自由意志的核心内涵包含实践理性与自主选择。自然人具有自由意志的论断是一种拟制,因此不应人为缩小犯罪主体范围。类人型AI实体的自主行为能力是摆脱了产品拥有人控制的自主选择能力,具有自由意志所要求的自主选择属性,其核心特质是具有深度学习能力,这是智能实体发展到类人阶段的标志,且符合实践理性的要求。类人型AI实体具有刑事可罚性,若不追究其自主实施犯罪行为的责任,就违背罪责自负原则。对其科处刑法处罚能够满足报应论与预防论的要求,应从体系上确立自由刑、修改程序及销毁实体等刑罚措施。

In order to properly protect legal rights and prevent crimes,it is necessary to carefully clarify the issue of whether AI entities can have the subject status of criminal responsibility.The academy has formed the affirmative theory,the negative theory and the terminative theory.However,most of the arguments show some disadvantages,such as generalization,confusion of concepts and one-sided negation.Normative shaping should be carried out through screening whether AI entities have free will and capability of criminal punishment.The core connotation of free will as a type consists practical rationality and autonomous choice.The assertion that natural individuals have free will is a fiction and should not artificially narrow the scope of the subject.The autonomous behavior capability of a human-like AI entity is the ability of autonomous choice that is free from the control of the product owner,and has the independent choice attribute required by free will.Its core characteristic is the ability of deep learning,which is a sign of the development of intelligent entity to the humanoid stage and meets the requirements of practical rationality.Humanoid AI entities are criminally punishable.If they do not hold accountable for their own criminal acts,they will violate the principle of self-responsibility.Punishing them with criminal law can meet the requirements of the theory of retribution and prevention,and the punishment measures such as free punishment,procedure modification and entity destruction should be established systematically.