The Value Conflicts And Countermeasures Of Algorithm Involved In Criminal Judgment Prediction

The progress of artificial intelligence technology boosts the integration of algorithm and forecast of criminal judgment.It is a necessary choice for algorithm to intervene in the forecast of criminal judgment against the background of judicial staffing system and the reform of lawsuit system focusing on trial.Deep learning algorithm and powerful computing power,massive criminal justice big data and element extraction technology,legal knowledge graph and forecast model construction,these aspects offer the basic conditions to the algorithm to realize the forecast of criminal judgment.Jointly by Shanghai high court and dispatch of i FLYTEK,for instance,"206" system of Shanghai has now been put into operation,which has the function of automatic identification details of a case,the review the evidence,the push and generation of the class case,and other functions.Thus the algorithm implementation on conviction and sentence on the criminal judgment as well as the court of intelligent construction have been pushed to a new height.From the dialectic point of view,everything has two sides.The aim of the algorithm interfered in the forecast of criminal judgment is to achieve the high efficiency and justice in criminal trial.For one thing,the algorithm involved in the forecast of criminal judgment can boost the trial efficiency and solve the problem of the number of cases with relatively few judge in our country.For another,the controversial and possibly unfair decision by COMPAS against Loomis shows that the algorithmic system is not so perfect.Since black box method and algorithm of discrimination is hard to avoid,the science and technology company leads the development of criminal judgment prediction algorithm simulation,the algorithm judges for interest measure and legal interpretation and so on,which leading to criminal judgment prediction algorithm in efficiency but the results may lack of justice.In order to achieve the goal of high efficiency and fairness in criminal trials,and to resolve the value conflicts caused by the involvement of algorithms in the prediction of criminal judgments,the innovative strategy and response idea of this article is to integrate existing methods and balance the value of judicial justice and judicial efficiency.First of all,the application boundary of the algorithm and the applicable case types should be determined,and the auxiliary and corrective functions of the algorithm should be used.Simple cases refer to the prediction results of the algorithm,and complex cases are at the discretion of the judge.Secondly,the algorithm counterpart should be given multiple procedural relief rights,including: the implementation of algorithm disclosure and limited transparency of the algorithm;the implementation of algorithm interpretation,giving the algorithm counterpart the right to request algorithm interpretation;the implementation of algorithm auditing and algorithm accountability and other algorithmic relief principles.Finally,accountability should be held for the consequences of the damage caused by the algorithm prediction results.Algorithm engineers are obliged to ensure the fairness of the algorithm system,and judges shall bear responsibility for faults in the algorithm prediction results they use.In short,in the field of criminal judgment prediction,it is necessary to develop and use algorithms on the basis of ensuring the fairness of the algorithm itself,so as to play the greatest positive role of the algorithm in the prediction of criminal judgments.