“预测正义”能否预测正义?——基于法国司法大数据预测应用的考察与启示

An Examination and Implications Based on the Predictive Application of Judicial Big Data in France

在着力发展司法大数据预测应用的当下,提升司法质量和效率的价值目标伴随着信息安全风险、算法偏见、决策独立性降低、解释性判例体系冲击等阴霾。面对相似的挑战,对具有成文法传统的中法两国司法大数据预测应用的比较研究具有现实意义。法国以2016年《数字共和国法》颁布为标志,推进以司法大数据预测应用为内容的“预测正义”。在强保护立场和保守基调下,法国预测正义在功能局限与价值冲突上备受质疑,司法个人信息保护、决策独立性、算法有效性、解释性判例体系嬗变等问题成为了争议的焦点。法国通过公开裁判信息预处理、“法官画像”禁令、以“等距法”理论更新司法范式、倡导异议裁判支持机制、提高算法透明度等途径,试图对冲预测正义带来的冲击。立足中国实际,批判借鉴法国经验教训,建议强化司法大数据中信息权的保护、确立司法智能决策有限参考原则、构建算法专业监督机制和异议裁判保护机制,以完善和发展我国人工智能辅助司法决策制度。

At the moment when efforts are being made to develop the predictive application of judicial big data,the goal of improving judicial quality and efficiency is accompanied by the haze of information security risks,algorithmic bias,reduced decision-making independence,and the impact of interpretive jurisprudence systems.In the face of similar challenges,a comparative study of judicial big data prediction applications in China and France,which have a statutory law tradition,is of practical significance.In France,the promulgation of the 2016 Digital Republic Act marked the promotion of"predictive justice"with the predictive application of judicial big data.Under the strong protection stance and conservative tone,France's predictive justice has been questioned in terms of functional limitations and value conflicts,and issues such as the protection of judicial personal information,decision-making independence,algorithmic validity,and the transmutation of interpretive jurisprudence system have become the focus of controversy.France has tried to hedge the impact of predictive justice by publicizing the pre-processing of adjudication information,banning the“judge's portrait”,updating the judicial paradigm with the theory of"isometric law",advocating the support mechanism of dissenting adjudication,and improving the transparency of the algorithm.Based on China's actual situation and drawing critically on the lessons learned from France,it is proposed that China's artificial intelligence-assisted judicial decision-making system be improved and developed in terms of strengthening the protection of the right to information in judicial big data,establishing the principle of limited reference for judicial intelligent decision-making,and constructing a mechanism for the professional supervision of algorithms and a mechanism for the protection of dissenting judgments.