首页|法律人工智能新范式:封闭与开放的二元兼容

法律人工智能新范式:封闭与开放的二元兼容

扫码查看
符号主义与联结主义两种经典人工智能范式,以及由此形成的基于规则和基于数据的法律推理模型,对应于规范封闭与认知开放这一古老的法理学命题.符号主义强调规范的封闭,联结主义强调认知的开放,两条路径代表了两种迥异的技术范式,在其背后潜藏着深刻的哲学思想差异,并导向两种完全不同的法律发展模式.法律人工智能的封闭与开放,因此涵盖了一系列重要的二元论法理学命题.可以围绕这组二元概念,在形式理性法和实质理性法的思想传统中,借助系统论法学的概念工具,重构人工智能法学的核心理论命题.大数据智能技术的单向驱动,将在法律的社会、时间和事实面向引发深刻的法律变异.这要求从法律的功能、领域、过程和结构层面,重新定位法律人工智能的三种演化类型.应当从规范封闭与认知开放的二元兼容视角,发展更为复杂的法律理论,以此设计法律人工智能的发展路线图,引导数字时代智能法律的良性发展.
Symbolism and connectionism are two classic paradigms of artificial intelligence,leading to rule-based and data-based legal reasoning models,respectively.These models correspond to the age-old jurisprudential proposition of normative closure and cognitive openness.Symbolism emphasizes the clo-sure of norms,while connectionism focuses on the openness of cognition.These two paths represent dis-tinct technological paradigms,underpinned by deep philosophical differences,leading to two entirely dif-ferent legal development models.Thus,the openness and closure of legal AI encompass a range of im-portant dualistic jurisprudential propositions.Around this dual concept,within the intellectual traditions of formal rational law and substantive rational law,using the conceptual tools of systems theory jurispru-dence,the core theoretical propositions of AI in law can be reconstructed.The unilateral drive of big data intelligence technology will provoke profound legal mutations across the social,temporal,and factual di-mensions of law.This demands a reevaluation of the functions,domains,processes,and structures of le-gal AI,defining three types of evolutionary paths.From the dual perspective of normative closure and cognitive openness,more complex legal theories should be developed to design the developmental road-map for legal AI,guiding the healthy evolution of intelligent law in the digital age.

Legal Artificial IntelligenceNormative ClosureCognitive OpennessPersonalized LawReflective Law

余成峰

展开 >

北京航空航天大学人文与社会科学高等研究院

法律人工智能 规范封闭 认知开放 个性化法律 反思法

国家社会科学基金重点项目

22AZD149

2024

中外法学
北京大学

中外法学

CSTPCDCSSCICHSSCD北大核心
影响因子:3.903
ISSN:1002-4875
年,卷(期):2024.36(3)
  • 8