人工智能时代法律推理的基本模式--基于可废止逻辑的刻画

从逻辑学的角度看,法律推理具有非单调性。人工智能时代更清晰地凸显出了与这一特性相应的可废止推理模式的必要性。可废止推理虽未必一定用可废止逻辑来刻画,但这一做法在人工智能的环境下更加合乎目的。法律推理的可废止性源于法律规则的可废止性,法律规则的逻辑形式化要求将其构成要件表征为"有待证明的要素"(P要素)与"未被驳倒的要素"(NR要素)两部分,后者的引入恰当地处理了规则与例外的关系。在此基础上,可以通过引入三类"废止者",即反驳型废止者、截断型废止者和削弱型废止者,来建构可废止法律推理的基本模型。但这同时也显现出了可废止法律推理的智能化限度,核心在于它无法进行司法裁判中必不可少的价值判断。

The Basic Model of Legal Reasoning in AI Age:Characterization Based on Defeasible Logic

Legal reasoning,seen from the perspective of logic,is nonmonotonic,and is typified by the defeasible model of reasoning,which has become ever more needed in an AI age.Though the defeasible reasoning may not necessarily be characterized in terms of defeasible logic,but such approach is more justified in the scenario of artificial intelligence.The defeasibility of legal reasoning originates from the defeasibility of legal rules,as the logical form of a legal rule is represented by two constituents,"the elements to be proved"(P-elements)and"the elements to be undefeated"(NR-elements).The introduction of NR-elements can properly deal with the relationship between rules and their exceptions.A further introduction of three kinds of"defeaters,"i.e.,rebutting defeater,undercutting defeater and undermining defeater will build a basic model of defeasible legal reasoning.But such an endeavor will also reveal the limits in the role that the AI can play in defeasible legal reasoning:the AI is incapable of making value judgments that are indispensable in judicial adjudication.