人工智能在防范刑事错案中的应用与制度设计

The Application and Development of Artificial Intelligence in Preventing Wrongful Conviction in Criminal Cases

针对因证据导致发生刑事错案的问题,司法机关积极探索将人工智能融入证据裁判过程,通过数据编码、文本——数据链接、标准化司法产品输出等阶段参与证据裁判过程,以实现刑事错案的有效防范。当前,人工智能在证据标准指引、证据校验及证据链审查判断等领域都有较多应用。但也面临着一些问题和挑战,如智能算法运行效率低下造成证据标准难以统一,以及智能算法运行数据偏差与隐秘导致刑事证据裁判的失正与失真等。因此,只有在实体上明确智能化决策模型的辅助功能,限定其应用范围,在程序上从启动方式、算法解释路径、算法评估与验证、救济程序等对智能化决策模型进行规范,才能真正建立起智能化的刑事错案防范体系。

In view of wrongful conviction in criminal cases arsing from evidence,the judiciary actively tries to involve artificial intelligence into the process of evidence adjudication.It can effectively prevent miscarriage of justice happened by inviting artificial intelligence to participate in the process of evidence adjudication through coding data,linking text and data,outputting standardized judicial product and other else.At present,artificial intelligence has been well applied in the fields of evidence guideline,evidence verification,reviewing chain of all evidence.However in the meantime,it also faces some difficulties and challenges,such as the low efficiency of the intelligent algorithm results in the difficulty of unifying the evidence standards,and the data deviation and privacy of the intelligent algorithm results in impartiality and inauthentic evidence.It is admitted that the preventing system of wrongful conviction through artificial intelligence is going to be created by defining substantively the auxiliary function of intelligent decision-making model and limiting its application scope,and regulating procedural ways of the intelligent decision-making model in program from starting mode,algorithm interpretation path,algorithm evaluation and verification,and relief program.