【摘要】 较之于裁判结果预测，我们更应关注司法人工智能的辅助说理能力。司法人工智能定位于裁判说理的辅助地位，在于算法的解释技术难以满足司法公开的要求，推理过程难以对接庭审的复杂辩论，推理的方法也难以改造成司法推理的方式，以上局限决定了司法人工智能难以独立完成针对裁判结果的说理任务。人机间更合适的定位应当是由法官承担说理责任，机器予以全面辅助。人工智能的辅助价值体现在为结论寻求充分的理由，集中于提升文书写作的水准。生成式大模型等新技术的出现，提升了司法人工智能辅助说理的能力，技术实现的方法具体包括实体方面提供更为充分的大小前提素材，形式方面提供文书写作的规范体例。更多还原

【Abstract】 Compared with the prediction of judgment results, we should pay more attention to the auxiliary reasoning ability of judicial artificial intelligence. The main reason is that the interpretation technology of the algorithm is difficult to meet the requirements of judicial publicity, the reasoning process is difficult to match the complex debate of the trial, and the reasoning method is also difficult to fully fit the synthesis of judicial reasoning. The above limitations determine that judicial artificial intelligence is difficult to independently complete the reasoning task for the judgment result. The reasonable position between man and machine should be up to judges to assume the responsibility of reasoning, and the machine should provide comprehensive assistance. The auxiliary value of artificial intelligence is that it can seek sufficient reasons for judgment conclusions and improve the standard of writing. The emergence of new technologies such as generative model has improved the ability of judicial artificial intelligence to assist reasoning. In terms of technology, the method includes providing more sufficient material of major premise and minor premise in terms of entity, and providing normative style of document writing in terms of form.更多还原