刑事审判中人工智能的进与退

The Function of Artificial Intelligence in Criminal Justice

作为辅助刑事审判的"新贵",人工智能须提供有益的智能。忧患主要有算法黑箱、类案推送与外包服务的长远性、人工智能分析结果的参考度及偏差率、网络安全威胁。人工智能在刑事审判中的"自由空间"是法律在预设空间内承认其运用的效力,具体包括协同促进、管理记录、文书初稿形成;"禁区"指法律不承认在这一空间其运用的效力,具体是预测定罪量刑结果;对证据能力进行实质判断;判断证据证明力以及闭合证据链;刑事案件疑难问题的价值判断。应当以坚持人工智能工具性作用最优化为总原则,要求其分析结果具有可解释性.增设刑事司法人工智能监管机构,建立问责制度与强制披露算法制度,全面升级法院内部网络的安全操作系统,促进全国法院刑事案件信息技术标准的逐步统一。

Artificial intelligence ought to provide useful intelligence to assist criminal trials, for which the main problems are the black box of the algorithm, the long-term delivery of categorized cases and outsourcing service, the reference degree and deviation rate of the artificial intelligence analysis result, and the network security threat. The "free space" of artificial intelligence in criminal trials is the effectiveness of the law in recognizing its application in the pre-set space, including the promotion of synergy, the management of records, and the formation of the first draft of the document. The uforbidden zone" means that the law does not recognize its force in this space, which specifically includes the prediction of the conviction and sentencing outcome, the substantive judgment of evidence ability, the judgment of evidence and closed evidence chain, and the value judgment of problems in criminal cases. With the general principle of optimizing the instrumental role of artificial intelligence, the analysis results of artificial intelligence should be interpretable, criminal justice artificial intelligence supervision institutions should be established, the artificial intelligence accountability system and mandatory disclosure algorithm system should be constructed, the security operating system of the court internal network should be upgraded, and the information technology standards of criminal cases in national courts should be gradually unified.