«上一篇/Previous Article|本期目录/Table of Contents|下一篇/Next Article»

《长安大学学报（社科版）》[ISSN:1671-6248/CN:61-1391/C]

卷:

期数:

2025年06期

页码:

120-132

栏目:

法学

出版日期:

2025-12-30

文章信息/Info

Title:

DeepSeek embedded in the formation process of judicial decisions:A new pathway for integrating intelligence technology with the rule-of-law paradigm

文章编号:

1671-6248(2025)06-0120-13

作者:

王译¹; 赵弋雅²

(1. 湘潭大学 法学学部,湖南 湘潭 411105; 2. 湘潭大学 纪检监察研究院,湖南 湘潭 411105)

Author(s):

WANG Yi¹;  ZHAO Yiya²

(1. Faculty of Law, Xiangtan University, Xiangtan 411105, Hunan, China; 2. Research Institute of Discipline Inspection and Supervision, Xiangtan University, Xiangtan 411105, Hunan, China)

关键词:

DeepSeek;  ChatGPT;  人工智能技术;  情报技术;  裁判文书;  算法解释

Keywords:

DeepSeek;  ChatGPT;  artificial intelligence technology;  intelligence technology;  judicial decision;  algorithm explanation

分类号:

D926; TP18

DOI:

-

文献标志码:

A

摘要:

为探索人工智能技术与裁判文书写作的深度融合路径,针对当前司法实践中存在的裁判文书质量参差不齐、撰写效率较低、标准不统一等现实困境,采用理论推演与实践验证相结合的方法,系统梳理裁判文书撰写过程中的关键痛点,结合DeepSeek的自然语言生成能力与司法实务需求,构建情报技术与法治融合的新范式。研究发现,DeepSeek在裁判文书写作中展现出显著的应用价值,该模型能够自动化完成事实归纳、法律条文检索等基础性工作,有效减轻法官的工作负担,大幅提升文书撰写效率,然而,该项技术的应用面临着司法数据安全保护、算法黑箱可能影响司法透明度、过度依赖技术可能削弱法官主体责任等多重挑战。研究表明,DeepSeek在裁判文书写作中的应用需要在技术赋能与风险控制之间建立动态平衡,为实现这一目标,应构建完善的司法数据分级保护体系,明确法官在智能辅助环境下的核心决策权边界,制定AI司法应用的伦理规范与行业标准,在确保技术应用合规性的同时,实现司法效能提升与风险治理的协同发展,为智慧司法建设提供可持续的技术支持路径。

Abstract:

To explore pathways for the in-depth integration of artificial intelligence technologies with judicial decision writing, and in response to practical challenges in current judicial practice-such as uneven quality of judicial decisions, low drafting efficiency, and non-uniform standards-this paper adopts a combined approach of theoretical deduction and practical validation. It systematically identifies key pain points in the process of drafting judicial decisions and, by aligning DeepSeek's natural language generation capabilities with the needs of judicial practice, constructs a new paradigm for the integration of intelligence technology and the rule of law. The findings reveal that DeepSeek demonstrates significant application value in judicial decision writing. Its model is capable of automating foundational tasks such as fact summarization and legal provision retrieval, effectively reducing judges' workloads and substantially improving drafting efficiency. However, the application of this technology also faces multiple challenges, including the protection of judicial data security, the potential impact of algorithmic “black boxes” on judicial transparency, and the risk that excessive reliance on technology may weaken the judges' primary responsibility. The study indicates that the application of DeepSeek in judicial decision writing requires establishing a dynamic balance between technological empowerment and risk control. To achieve this goal, it is necessary to build a comprehensive tiered protection system for judicial data, clearly define the boundaries of judges' core decision-making authority in intelligence-assisted environments, and formulate ethical norms and industry standards for AI applications in the judiciary field. By ensuring the compliance of technological applications, coordinated progress can be achieved in enhancing judicial efficiency and managing risks, thereby providing a sustainable technological support pathway for the development of smart justice.

参考文献/References:

[1] 吉林高院新闻办.24小时自助服务:运用“智慧法院”跑出诉讼“加速度”[EB/OL].(2022-11-15)[2025-06-11].http://jlfy.e-court.gov.cn/article/detail/2022/11/id/7016396.shtml.
[2]蔡天琪,蔡恒进.DeepSeek的技术创新与生成式AI的能力上限[J].新疆师范大学学报(哲学社会科学版),2025(4):136-143.
[3]叶继元,成颖.情报的概念及其与信息链、DIKW链的关系探讨[J].中国图书馆学报,2022(4):39-51.
[4]赵精武,何傲翾.“智能化锦标赛”下数字政府治理理论与治理模式创新——基于“DeepSeek+政务”实践的分析[J].电子政务,2025(6):16-30.
[5]彭中礼.现代科技赋能司法现代化的逻辑机理及其实践路径[J].中共中央党校(国家行政学院)学报,2025(2):131-142.
[6]刘瑜.ChatGPT类人工智能赋能事实认定的前景展望[J].司法改革论评,2023(2):56-75.
[7]卢荣婕.ChatGPT赋能智慧法院建设:机遇、挑战和规制[J].重庆邮电大学学报(社会科学版),2024(3):83-93.
[8]张欣.生成式人工智能的数据风险与治理路径[J].法律科学(西北政法大学学报),2023(5):42-54.
[9]杨帆,吕士哲.智慧司法建设下审判主体的异化风险及防范[J].中南民族大学学报(人文社会科学版),2025(1):109-117,185-186.
[10]宋华健.论生成式人工智能的法律风险与治理路径[J].北京理工大学学报(社会科学版),2024(3):134-143.
[11]PATEL D, WONG G. GPT-4 architecture, infrastructure, training dataset, costs, vision, moe[EB/OL].(2023-07-10)[2025-05-30]. https://semianalysis.com/2023/07/10/gpt-4-architecture-infrastructure/?nthPub=11.
[12]LIU A,FENG B,Xue B,et al.DeepSeek-V3 technical report[EB/OL].(2025-02-18)[2025-05-21].https://arxiv.org/pdf/2412.19437v1.
[13]张渊.ChatGPT模型文本的生成机制与文本特性分析[J].重庆文理学院学报(社会科学版),2024(4):105-114.
[14]赵葛剑,张新鹏.DeepSeek:从“概率生成”到“因果推理”[J].自然杂志,2025(2):79-84.
[15]上海金融法院.上海金融法院发布《债券纠纷法律风险防范报告》[EB/OL].(2023-03-02)[2025-05-25].https://www.hshfy.sh.cn/shfy/web/xxnr.jsp?pa=aaWQ9MTAyMDI5MTk3MiZ4aD0xJmxtZG09bG0xNzEPdcssz.
[16]郑曦.刑事司法中的数据安全保护问题研究[J].东方法学,2021(5):80-92.
[17]康京涛,王砼.公共数据开放的安全风险及应对策略[J].情报杂志,2024(9):166-171.
[18]肖冬梅,苏莹.中国政府数据开放中的安全风险及其防范对策[J].现代情报,2022(6):112-120,131.
[19]沈阳,闫佳琦,邹开元.基于DeepSeek思维链的人机协同知识生产[J].中国编辑,2025(4):95-100.
[20]ZHENG R,DOU S,GAO S,et al.Secrets of RLHF in large language models part I:PPO[EB/OL].(2023-07-18)[2025-05-25].https://arxiv.org/pdf/2307.04964.
[21]刘友华.算法偏见及其规制路径研究[J].法学杂志,2019(6):55-66.
[22]夏雨.算法鸿沟的法律弥合[J].天水师范学院学报,2024(4):2-10.
[23]胡键.“算法依赖症”及其对国家治理现代化的影响[J].国家现代化建设研究,2023(4):135-148.
[24]DILLON S.The Eliza effect and its dangers:from demystification to gender critique[J].Journal for cultural research,2020(1):1-15.
[25]BAUSERMAN V.Unemployment insurance agency[EB/OL].(2024-11-12)[2025-05-25].https://law.justia.com/cases/michigan/supreme-court/2019/156389.html.
[26]衣俊霖.数字孪生时代的法律与问责——通过技术标准透视算法黑箱[J].东方法学,2021(4):77-92.
[27]TORNATZKY L G,FLEISCHER M.The processes of technological innovation[M].Lexington:Lexington Books,1990.
[28]邹纯龙,马海群,王今.公共数据安全韧性治理体系的系统动力学分析[J].图书情报工作,2024(3):3-14.
[29]黄伟庆.合规视角下数据要素的分类分级管理机制研究[J].上海政法学院学报,2024(2):121-140.
[30]苏志明.区块链技术在保密通信中的应用[J].电子技术与软件工程,2019(24):27-28.
[31]吴婧.中央首次部署公共数据资源开发数据领域发展大有可为[N].中国经营报,2024-10-14(4).

相似文献/References:

[1]张夏恒.基于新一代人工智能技术(ChatGPT)的数字经济发展研究[J].长安大学学报(社科版),2023,25(03):55.
　ZHANG Xiaheng.Research on the development of digital economy based on the new generation of artificial intelligence technology(ChatGPT)[J].Journal of Chang＇an University(Social Science Edition),2023,25(06):55.

备注/Memo

备注/Memo:

收稿日期:2025-06-16
基金项目:国家重点研发计划重点专项项目(2020YFC0832400); 湖南省社会科学基金项目(23YBQ055); 湖南省教育厅党的二十大精神研究专项项目(23B1142); 湖南省大学生创新训练计划项目(S202410530188)
作者简介:王译(1988-),男,湖南邵阳人,副教授,法学博士。

常用功能

更新日期/Last Update: 2025-12-30