首页|算法辅助法律解释的规范性构建

算法辅助法律解释的规范性构建

扫码查看
算法技术的迅猛发展使算法辅助法律解释成为可能,相较于传统之法律解释方法,算法辅助法律解释可演进出以读取规则存在意义为核心的感知解释、以展示规则后果实效为核心的动态解释以及以理解规则关联反思为核心的归约解释.算法辅助法律解释之规范性构建,同样面临新法律解释方法间的效力冲突问题.在方法运用的选择上,感知解释中数据运算与规范价值间的张力有可能倒向轻率决断;动态解释中预测模型的过度使用存在技术主义至上的隐患;归约解释亦需要避免法律解释中法律原则与法律规则的适用争议.算法辅助法律解释规范性的实现,需要融贯考量算法辅助法律解释各方法的使用限度.在司法实践场景中应优先使用感知解释方法,完善理性司法下的技术供给;动态解释着重于进行后果考量因而应当次于感知解释使用;当感知解释与动态解释的使用面临艰难抉择时,应以归约解释作为调和工具,寻得个案中法律解释者对解释方法使用的共识.
Normative Construction of Algorithm Aided Legal Interpretation
The rapid development of artificial intelligence technology and algorithmic technology makes algorithm ai-ded legal interpretation possible. Following the idea of drawing the form from the normative text of legal provisions to the symbolic language that can be processed,handled and applied by algorithms,introducing Peirce' s symbolic theoretical system,as well as referring to the computer modal language,the algorithm aided legal interpretation,compared with the traditional legal interpretation,can evolve into the perceptual interpretation centered on the read-ing of the meaning of the rules,the dynamic interpretation centered on the demonstration of the consequences of the rules,and the reductive interpretation centered on the reflection of the understanding of the association of the rules. The normative construction of algorithm aided legal interpretation also faces the problem of conflict of effectiveness among new legal interpretation methods. In the choice of method,the tension between data operation and normative value in perceptual interpretation may lead to rash decision,the excessive use of predictive models in dynamic in-terpretation may potentially pose the hidden danger of technocratic supremacy,and reductive interpretation also needs to avoid the controversy between the application of legal principles and legal rules in legal interpretation. The realization of the normative nature of algorithm aided legal interpretation requires a coherent consideration of the limits of the use of various methods of algorithm aided legal interpretation. The judiciary should constrain the use of algorithms in legal interpretation with human rationality,and establish a priority level of"perceptual interpretation-dynamic interpretation-reductive interpretation"with human rationality,to promote the legal interpreters to show their rational ability while obtaining technical support and optimize the procedures and results of algorithm aided le-gal interpretation. In judicial practice,the use of perceptual interpretation method should be prioritized to improve the rational chain of"algorithmic assistance-human decision",so that if the interpretation result supported by suffi-cient reasons can be obtained through perceptual interpretation,and meets the requirements of meaning,truth value and normative modality,its interpretation conclusion can be applied with reference to the common ideas of textual interpretation. Dynamic interpretation focuses on the consideration of consequences and should be used second to perceptual interpretation. The interpreters should limit the dynamic interpretation to the scope of consequence sup-plementation based on the distinction between legal interpretation and legal prediction. When the use of perceptual interpretation and dynamic interpretation is faced with a difficult choice,reductive interpretation should be used as a tool for reconciliation to find consensus among legal interpreters on the use of interpretation methods in individual cases,forming the algorithmic conditions of"legal norms-rules of operation". It is precisely the"human legisla-tion"that establishes the reference status and instrumental status of the algorithm aided legal interpretation,and al-so provides the most fundamental and core compliance for the normative construction of the algorithm aided legal in-terpretation.

judicial artificial intelligencealgorithm aided legal interpretationtextual interpretationconsequence argument

费凯学

展开 >

伊犁师范大学 法学院,新疆 伊宁 835000

司法人工智能 算法辅助法律解释 文义解释 后果论证

2024

中国海商法研究
中国海商法协会 大连海事大学

中国海商法研究

CSTPCDCHSSCD
影响因子:0.391
ISSN:1003-7659
年,卷(期):2024.35(4)