智慧司法场域下人工智能辅助精准量刑的理论供给与规范应用——以法官行使刑事自由裁量权为切入点

Theoretical Supply and Normative Application of AI-Assisted Precision Sentencing from the Perspective of Smart Justice—Focusing on the Exercise of Criminal Discretion by Judges

摘要: 面对“案多人少”与“类案不同判”的双重压力，经验式量刑已难以满足公正与效率并重的司法需求。研究以规范分析与技术法学交叉的方法，聚焦人工智能介入量刑之理论与实践，提出以“数据–规则–算法”三维映射重塑法院裁量结构：在责任刑–预防刑区分框架下，机器完成量化计算，人类保留价值判断，实现“量化可以量化的、裁量必须裁量的”权责配置。进而研究揭示数据偏差、算法黑箱、结果依赖与信息安全四重风险对实体公正和程序正当性的系统性冲击，对此针对性地提出四项规范化路径。最终研究指出，人工智能应定位为裁量支持而非裁量主体，唯有在技术理性与司法理性深度耦合、实体与程序并重的前提下方能真正实现精准量刑与“同案同判”，为我国智慧司法提供可复制、可推广的制度蓝本。

Abstract: Facing the dual pressures of heavy caseloads with insufficient judicial resources and “different judgments for similar cases”, experience-based sentencing has struggled to meet the judicial needs of balancing fairness and efficiency. This study employs a cross-methodology of normative analysis and technical jurisprudence, focusing on the theory and practice of artificial intelligence (AI) intervention in sentencing. It proposes a three-dimensional mapping of “data-rules-algorithms” to reshape the court’s discretionary structure: under the framework distinguishing between liability punishment and preventive punishment, machines complete quantitative calculations, while humans retain value judgments, realizing the power-responsibility allocation of “quantifying what can be quantified and exercising discretion where necessary”. The study further reveals that four risks-data bias, algorithmic black box, result dependency, and information security-systematically impact substantive justice and procedural legitimacy. In response, four standardized pathways are proposed. The study finally emphasizes that AI should be positioned as discretionary support rather than the discretionary subject. Only under the premise of deep integration between technical rationality and judicial rationality, and equal emphasis on substance and procedure, can precision sentencing and “same cases, same judgments” be truly achieved, providing a reproducible and promotable institutional blueprint for China’s smart justice.