From ‘Efficiency’ to ‘Justice’: Cross-domain Translation Distortion of the Concept of Algorithm in Judicial Contexts and Its Regulation

Abstract

With the widespread application of artificial intelligence in the judicial field, the algorithm, a core concept of computer science, is being translated across domains into the institutional context of adjudication. Drawing on the three-layer framework of conceptual structure–the propositional kernel, contextual activation frame, and institutional embedding structure–and the RID model in Knowing and Saying: An Ontological Investigation of Human Cognition , this paper analyzes the distortion that occurs when the concept of algorithm enters judicial practice. The study argues that the algorithmic kernel of optimization is metaphorically generalized when it is mapped onto the judicial goal of justice. At the contextual level, the engineering pursuit of efficiency and consistency conflicts with the judicial pursuit of individual justice and procedural participation. At the institutional level, the black-box character of commercial algorithms dislocates technical responsibility and judicial responsibility. Through a comparative analysis of the COMPAS risk assessment system in the United States and sentencing-assistance systems in China’s Smart Court program, including the 206 System, the paper reveals heterogeneous forms of translation distortion under different judicial systems. It concludes that a productive interaction between algorithms and judicial justice requires litigation-oriented algorithmic review, the dismantling of algorithmic black boxes, and the reconstruction of judicial responsibility in the age of artificial intelligence.