人工智能刑法"主体性"否定:缘起、解构、反思——以认知科学的五个层级为基础
CSTR:
作者:
作者单位:

作者简介:

通讯作者:

中图分类号:

D924.13;TP18

基金项目:

中国博士后科学基金第71批面上资助"智慧量刑的悖论与破解研究"(2022M712650);2019年度西南政法大学科研创新项目"人工智能的刑法属性研究"(FXY2019009)


The negation of the "subjectivity" of artificial intelligence in criminal law: Origins, deconstruction and reflection:based on the five-level theory of cognitive science
Author:
Affiliation:

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    人工智能是否能与人类一样具有刑法"主体性"地位无法在刑法理论内部找到答案,而需要以相关的认知科学为基础,否则就会陷入循环论证的困境。根据认知科学的五个层级理论,人工智能在神经层级、心理层级的低阶认知层面,仅是对人类认知的简单模拟;在作为高阶认知与低阶认知中间环节的语言层级认知层面,人工智能的人工语言与人类自然语言之间具有本质的区别;在思维层级、文化层级的高阶认知层面,当前的人工智能并没有显现出其具有思维或产生文化的能力。结合塞尔"中文房间模型"研判,人工智能并不具备刑法意义的"辨认能力"与"控制能力"。在辨认能力方面,人工智能传感器对客观世界的识别虽然能在一定程度上对人类认知进行形式模拟,但其并不能将识别到的信息与自身"行为"结合进行加工处理从而得出该"行为"的社会意义;在控制能力方面,人工智能所展示出的"控制能力"实质上是一种执行算法的能力,其本质上仍然是一种人类控制,而非人工智能的"自我控制"。因此,在当下及可预见的未来,人工智能并不具有刑法"主体性",进而承担刑事责任,当前的刑法理论也不必对所谓的"强人工智能"过度反应。将"类人"的人工智能提升到人类同等高度,有损人之所以为人的尊严。将"删除数据、修改程序、永久销毁"等与刑罚异质的要素纳入刑法会让其有适用于人类的危险。故在当下及可预见的未来,人工智能对刑法理论的影响主要在于其导致传统社会风险加剧,刑法理论的应对模式仍应当在现有刑法理论体系内,结合风险刑法理论将其作为"犯罪对象"和"犯罪工具"对待。当人工智能作为犯罪对象时,其是以财物、作品等形式存在,在司法认定过程中要注意界分人工智能本身与人工智能的载体;当其作为犯罪工具时,会导致犯罪后果扩大,查证难度增大等结果。

    Abstract:

    Whether artificial intelligence can have the same "subjectivity" status as human beings in criminal law cannot be found in the criminal law theory, and it needs to be based on relevant cognitive science, otherwise it will fall into the dilemma of circular argument. According to the five-level theory of cognitive science, AI is only a simple simulation of human cognition at the lower-order cognitive levels of the neural and mental hierarchies; At the level of language-level cognition, which is the intermediate link between higher-order and lower-order cognition, there is an essential difference between the artificial language of AI and natural language; At the higher-order cognitive levels of the thinking and cultural hierarchies, current AI has not shown itself to be capable of thinking or generating culture. Combined with the "the Chinese room argument", the AI does not have the "recognition ability" and "control ability" in criminal law theory. In terms of recognition ability, although the recognition of the objective world by AI sensors can mimic human cognition, they are not able to process the recognized information in conjunction with their own "behaviors" to derive the social significance of those "behaviors". In terms of control capability, the "control capability" demonstrated by AI is essentially the ability to execute algorithms, which is still essentially a form of human control rather than AI "self-control". Therefore, in the present and foreseeable future, AI does not have a "subjectivity" of criminal law, and current criminal law theory need not overreact to so-called "strong artificial intelligence". Elevating AI to human heights undermines human dignity. "Delete data, modify procedures, permanent destruction" is not a penalty. If it is written into the criminal law also has the possibility of human application. Therefore, in the present and foreseeable future, the impact of AI on the theory of criminal law is mainly due to the fact that it leads to the exacerbation of traditional societal risks. Criminal law theory should treat them as "objects" and "instruments" of crime in conjunction with the theory of risk criminal law. When AI is used as an object of crime, it exists in the form of property, works, etc., and care should be taken to distinguish between the AI itself and the carrier of the AI in the process of judicial determination; When it is used as an instrument of crime, it can lead to results such as widening the consequences of the crime and making it more difficult to investigate.

    参考文献
    相似文献
    引证文献
引用本文

甄航.人工智能刑法"主体性"否定:缘起、解构、反思——以认知科学的五个层级为基础[J].重庆大学学报社会科学版,2024,30(3):242-252. DOI:10.11835/j. issn.1008-5831. fx.2019.12.007

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:
  • 最后修改日期:
  • 录用日期:
  • 在线发布日期: 2024-07-07
  • 出版日期:
文章二维码