Research on the Procedural Justification Regulation of Sentencing Recommendations under the Human-Machine Collaboration Model

Abstract

The application of artificial intelligence (AI) technology in prosecutorial case handling is deepening, and sentencing recommendation assistance has become an important part of smart prosecutorial construction. AI provides data support and reference for prosecutors to propose sentencing recommendations through detailed analysis of rules such as elements of a crime and sentencing circumstances, promoting the transformation of sentencing activities from experience-driven to data- and experience-driven collaborative processes. However, the deep involvement of intelligent sentencing assistance has also raised issues such as algorithmic black boxes, leading to opaque decision-making processes, algorithmic bias, and potential procedural legitimacy risks such as systemic injustice, unclear allocation of human-machine decision-making weights, and the erosion of parties' procedural rights. This paper uses the principle of procedural legitimacy as an analytical framework, exploring the procedural regulation path of sentencing recommendations under the human-machine collaborative model from three dimensions: ensuring the substantive procedural participation of parties in the results of algorithm-assisted sentencing, constructing an explainable and auditable mechanism for algorithm-assisted sentencing recommendations, and clarifying the allocation of judicial responsibility for human-machine collaborative sentencing recommendations.