法律人工智慧的法哲學反省：判斷權限、執法機制與法治理念

Rule by Algorithms? A Preliminary Study on the Tension Between the Rule of Law and AI

法律人工智慧系統(ALI)對於法治價值的影響與關係尚無深入探討。本文主張ALI與法治具有內在的緊張關係。第一，法治理念中的「主體」是一種主動且負責任的主體概念觀，而日趨完善的ALI將凸顯出權限分配移轉的難題。第二，高效率執法機制的建置是ALI的核心特色，旨在以執法效率提升法律之應然與實然面向的擬合，然其卻可能損害個人自主的道德判斷空間與干擾行動者行動理據的排序。據此，本文從法哲學角度說明ALI與不同法治觀的可能緊張關係、透過因果圖呈現ALI對於守法行動的可能介入方式，進而提出兩個基本規範性原則：第一，替代型智慧系統之禁止原則。第二，高效率執法機制禁止原則。

Artificial legal intelligent systems have been extensively developed to perform various legal tasks. However, rare discussions focus on the question of whether such development might conflict with the idea of the rule of law. In this article I argue that there is an internal tension between ALIs and the rule of law based on two reasons. First, the idea of a legal subject in the rule of law is conceived as an active, responsible, and answerable agent who has the authority for making a decision. The fact that ALIs become more capable of performing legal tasks eventually leads to the possibility of shifting the decision-making authority away from human. The rule of law draws a baseline for developing ALIs, that is, ALIs should be supplementary, not substitute for human. Secondly, the highly efficient mechanism of law enforcement, one of the core fea-tures in ALIs, severely decreases an individual’s autonomous sphere of moral reasoning for ALIs seek for an assurance that an agent would comply with law through the certainty of legal sanctions. As a result of adopting this mechanism, ALIs aim at influencing an individual’s auton-omy of ranking reasons. Based on the two arguments, I demonstrate there are two principle on the regulation of ALIs. The first principle prohibits the development of substitute ALI. The second principle pro-hibits the widely adoption of highly efficient mechanism of law enforce-ment. This article examines the fundamental issue of the rule of law in the age of AI from the perspective of legal philosophy.