中国审批：人工智能让司法更加公正高效

“Artificial Intelligence” Makes Judicature More Just, Efficient and Authoritative–the Theoretical Analysis and Practical Exploration of “Artificial Intelligence” in Judicial Field.

2017年10月

Cui Yadong（崔亚东）, President and Chief Justice of Shanghai High People`s Court

October, 2017

2017年最热门的科技名词当“人工智能”莫属，这个已经存在了60年的技术领域因为谷歌的AlphaGo人机大战而声名鹊起。从过去的高不可及到今天的人人皆知，人工智能已经无处不在。Apple的Siri、蚂蚁金服的刷脸支付、Google的无人驾驶车等都有人工智能技术的身影。可以说，人工智能时代已经来临，人工智能的未来可能像工业革命、互联网一样再次给人类社会带来颠覆性的改变。[1]在司法领域，“人工智能+”已成为趋势。目前，上海法院积极推进人工智能等高科技应用与司法实践深度融合，推动大数据、“互联网+”到“人工智能+”的新跨越，促进上海法院审判体系和审判能力现代化实践的同时，形成了人工智能应用于司法领域理论与实践的认识与思考。

The most popular technology term in 2017 is “artificial intelligence”. This technology field that has been around for 60 years rose to fame because of Google AlphaGo’s battle with human. Being unreachable in the past, it is now a household word and is present everywhere, such as Apple’s Siri, Ant Financial’s face payment and Google’s driverless car and so on. It can be said that the era of artificial intelligence has come, and the future of artificial intelligence may bring revolutionary changes to human society ¹, like the industrial revolution and the Internet. In the field of justice, “artificial intelligence +” has become a trend. At present, Shanghai High People`s Court actively promotes the fusion of high-tech applications (such as artificial intelligence) and the judicial practice, driving a new stride from big data and “Internet +” to “artificial intelligence”, which help modernize Shanghai court judicial system and competence, and meanwhile conclude theoretical and practical cognition and thinking of artificial intelligence applied in judicial field.

一、   人工智能引入司法领域的理论认识与分析

Theoretical Understanding and Analysis of the Introduction of Artificial Intelligence in Judicial Field

（一）                 法律形式主义为人工智能司法应用提供了理论基础

Legal Formalism Provides Theoretical Basis for Artificial Intelligence’s Judicial Application

法律形式主义肇始于古罗马，是一种凸显逻辑作用的法律思想，其核心主张就是坚信法律制度是一个封闭的逻辑自足的概念体系，遵循三段论的推理逻辑模式，即以法律规范为大前提，以具体的案件事实为小前提，进而推导出裁判结果。[2]根据这一理论，机器只要遵守法律推理的逻辑，就可以得出公正的裁判结果。尽管这一理论受到诸如“自动售货机”现象[3]的批判，但从人工智能就是为思维提供机械论解释的意义上来说，法律形式主义对法律推理所作的机械论解释，恰恰为人工智能的司法应用提供了可能的理论前提。[4]

Originating in ancient Rome, legal formalism is a kind of legal thought that highlights logic function. Its core proposition is convinced that legal system is a closed system self-sufficient in logic. It keeps to syllogism reasoning logic model. In other words, it sets legal norm as the major premise and specific case facts as minor premise, and then deduces judgment. 2 According to this theory, a machine can draw a fair verdict by obeying the logic of legal reasoning. Although this theory is criticized as a “vending machine”³, from the perspective of artificial intelligence providing mechanistic explanation for thinking, legal formalism’s mechanistic explanation for legal reasoning just provides a possible theoretical premise for the judicial application of artificial intelligence.

（二）法律现实主义为推动人工智能模拟法官思维提供了理论支撑

Legal Realism Provides Theoretical Support for Artificial Intelligence Simulating Judge’s Thinking

法律现实主义是美国二十世纪上半叶兴起的一场法律变革的运动，从反面对传统法律方法提出质疑，倡导法律方法必须把社会利益的衡量引入到规范的法律论证当中。霍姆斯法官著名的格言“法律的生命并不在于逻辑而在于经验”[5]。法律现实主义对法官主观能动性和法律推理灵活性的强调，促使人工智能研究从模拟法律推理的外在逻辑形式进一步转向总结裁判经验中的规律性和普适性问题，探求法官的内在思维结构。[6]

Legal realism is a legal movement that rose in the 20th century in U.S.A, which questioned traditional legal method from a reverse side, and advocated that legal method must take social interest into the standard of legal argumentation. American jurist Oliver Wendell Holmes Jr.’s famous maxim “The life of the law has not been logic; it has been experience” 5. Legal realism’s emphasis on subjective judge and legal reasoning flexibility promotes artificial intelligence research changing from simulation of the external logic form of legal reasoning to conclusion of the regularity and universality of judgment and exploration of the inner thinking structure of the judge. 6

（三）“开放结构”理论为人工智能在司法领域的深度应用提供了理论创新

“Open Texture” Theory Provides Theoretical Innovation for Deep Application of Artificial Intelligence in Judicial Field

法律形式主义过于机械，而法律现实主义又过于自由，会殃及法治要求实现规则治理的根本原则，并动摇人工智能在司法领域存在的基础。因此，折中的“开放结构”理论，[7]既承认逻辑的局限性又强调其重要性；既否认法官完全按自己的直觉经验来随意判案的见解，又承认心证的存在。在这一理论的指导下，人工智能在司法领域可以进行更具深度和广度的应用，一方面是将简易问题从疑难问题中筛选出来，运用基于规则的技术来解决；另一方面是将疑难问题先用非案例知识，如规则、控辩双方的陈述、社会常识来获得初步答案，再运用案例来比对，检查案件的正确性。“开放结构”理论既肯定了法律的形式理性，又维护了司法实践不断发展的旺盛生命力，使人工智能对优化法官裁判科学性、准确性成为可能。上海高院承担研发“推进以审判为中心的诉讼制度改革软件”的任务正是对这一理论最好的实践和诠释。

Legal formalism is too mechanical, and legal realism is too free, which will adversely affect the fundamental principle of implementing rule governance required by rule of law, and shake the foundation of artificial intelligence in the judicial field. Therefore, this eclectic “Open Texture”7 recognizes the limitations of logic and emphasizes its importance, and denies that the judge judges the case completely based on his own intuitive experience and acknowledges the existence of judicial discretion. Under the guidance of this theory, artificial intelligence can have deeper and wider application in judicial field. On the one hand, rules are used to solve simple and easy questions screened out of complicated problems. On the other hand, it is necessary to use the non-case knowledge, such as the rules, the statements of two parties and common sense, to obtain the preliminary answers, and then compare different cases to check the correctness of the case. “Open Texture” theory affirms the ration of legal formalism and maintains the continuous developmental vitality of judicial practice, making it possible for artificial intelligence to optimize the scientificity and accuracy of judge’s judgment. Shanghai High People’s Court’s undertaking the task of “promoting the reform software of litigation system centered on court trial” is the best practice and interpretation of this theory.

二、   人工智能在我国司法应用的现状与实践价值

The Present Situation and Practical Value of Artificial Intelligence’s Application in China Justice

2017年7月9日，习近平总书记作出重要指示：“要遵循司法规律，把深化司法体制改革和现代科技应用结合起来，不断完善和发展中国特色社会主义司法制度。”

On July 9, 2017, General Secretary Xi Jinping made an important instruction: “we must follow the law of justice, and combine the task of deepening the reform of the judicial system and modern science and technology to develop and perfect the legal system of socialism with Chinese characteristics.”

人工智能新时代，我们不仅站在“巨人”的肩膀上，而且站在人类的“智慧之巅”。可以说，人工智能在司法领域的深度应用，提升了司法质量、司法效率、司法公信力，助力司法体制改革，助力法治中国建设，其价值与意义十分重大。上海法院的实践充分体现了其价值与意义。

In the new era of artificial intelligence, we not only stand on the shoulders of giants, but also on the top of human wisdom. Deep application of artificial intelligence in judicial field, so to speak, enjoys significant value in improving the quality of justice, judicial efficiency and judicial credibility, and boosting Chinese judicial system reform and Chinese construction of rule of law. The practice of Shanghai High People’s Court fully embodies its value and significance.

（一）    辅助司法办案，提升司法质效

Assist Judicial Handling and Improve Judicial Quality and Efficiency

不论是在判例法国家，还是成文法国家，浩如烟海的判例案卷以及各种法律法规、规章制度和司法解释，需要耗费司法人员大量的时间和精力去检索整理、分析筛选，而人的大脑认知和记忆能力是有局限性的。人工智能可以以其强大的存储、检索和分析功能，弥补人类智能和精力的有限性，极大地解放司法人员的重复脑力劳动，使其能够集中更多的精力、时间从事更加复杂的法律推理、侦查判断、自由裁量等，大大提升司法质量和效率。如上海法院建立了由智能辅助办案系统、智能终端办案APP、智能庭审系统、智能管理系统等35个子系统组成的“上海法院大数据审判辅助体系”，为法官办案提供了多元化、全覆盖、便捷式的智能化服务，日均最大访问量达1.8万余人次，已成为法官办案离不开的助手。

Whether in a Case Law or Statute Law country, voluminous case files and various laws, regulations and judicial interpretations require a lot of time and energy to retrieve, sort, analyze and screen while people’s brain cognitive and memory ability is limited. With powerful storage, retrieval and analysis function, artificial intelligence can make up human limitation of intelligence and energy. It can greatly liberate judicial official’s repetitive mental work and allow them to focus more energy and time on more complex legal reasoning, investigation, judgment and discretion, etc., which greatly improves the quality and efficiency of justice. For instance, courts in Shanghai established a “Shanghai court data system for trial assistance” which is made by 35 subsystems, such as an intelligent handling system, an APP of intelligent terminal handling, an intelligent hearing system and intelligent management system and so on, which provides the judge with diverse, comprehensive and convenient intelligent service. With a daily maximum visit of more than 18000 times, it has become a necessary assistant for the judge.

（二）    促进司法公正，提升司法公信

Promote Judicial Impartiality and Credibility

由于司法人员是具有主观能动性的差异个体，在执法统一标准时必然会产生一些差异，进而产生执法不统一、同案不同判等现象。利用人工智能，可以为司法审判提供相对统一的推理和评价标准，使法官能得到类似案件的全部案例以及法律、法规、司法解释等裁判规则，严格遵循证据规则、程序规则作出裁判，减少司法的任意性，有效防范冤假错案的产生，促进司法公正。同时，由于人工智能没有物质欲望和感情牵绊，不会受到外界的干扰和侵蚀，所以它能较好的排除人为的因素，最大限度地减少某些徇私舞弊、不公正的司法现象，提升司法公信力。如上海法院智能辅助办案系统具有类案推送，证据标准、证据规则指引，量刑参考、法条推送等功能，根据在办案件的特征，为办案人员自动推送同类案例，自动匹配程序性和实体性法律规范，对确保案件办案质量发挥了很大作用。

Because the judicial personnel are different individuals with subjective initiative, there will inevitably be some differences in enforcing uniform standard of law, which will result in inconsistent law enforcement and different judgment of the same case. Application of artificial intelligence can provide relatively unified judicial reasoning and evaluation standard, provide the judge with all similar cases, laws, regulations and judicial interpretations and so on, so the judge can strictly follow the rule of evidence and procedure, which will reduce judicial arbitrariness and effectively guard against unjust, false and erroneous cases, promoting the judicial justice. At the same time, due to artificial intelligence without material desires and feelings, it will not be subject to external disturbance and erosion, so it can eliminate human factors and minimize some practice favoritism and unfair judicial phenomenon, which will surely improve the judicial credibility. Shanghai court intelligent auxiliary handling system has many practical functions, like similar case push service, evidence standard, guidance on rule of evidence, sentence reference, articles of law push. It can automatically push the similar case, match procedural and substantive legal norms for the personnel according to the characteristics the case, which plays a important role in ensuring the quality of case handling.

（三）    服务群众诉讼，提升服务质量

Serve Public Litigation Needs and Improve Service Quality

司法为民是人民法院的根本宗旨，而高科技是践行这一根本宗旨的重要手段。上海法院在“互联网+”的基础上，将“人工智能+”植入12368诉讼服务平台，运用语音识别、自然语言理解、语音合成等人工智能技术，与诉讼服务深度融合，为人民群众提供了“全方位、全天候、零距离、无障碍”的诉讼服务,有效解决了人民群众反映强烈的案件查询难、诉讼咨询难、联系法官难、等诉讼难题。又如上海法院运用“互联网+”、“人工智能+”，建立了“上海法院律师服务平台”，向全国律师开放，为律师提供网上查询、网上立案、网上阅卷等诉讼服务，律师足不出户，即可完成诉讼事务，极大地方便了律师执业，保障了律师权益。

“Justice for people” is the fundamental tenet of the people’s court, and high technology is an important means to realize it. On the basis of the “Internet +”, Shanghai court equips 12368 hotline lawsuit service platform with artificial intelligence. Using speech recognition, natural language understanding and speech synthesis, it is highly fused with litigation services to provide people with “all-round, all-weather, zero distance, barrier-free” litigation service, which can effectively solve problems that people reflects strongly: difficulties in case query, litigation consultancy, contacting the judge and waiting for litigation results. Shanghai court use “Internet +”, and “artificial intelligence +” to establish the “Shanghai court lawyer service platform that is open to lawyers nationwide and provides online query, online case registration, online case files and other litigation services. Lawyers can complete litigation procedure without leaving home, which brings great convenience to lawyer, and safeguards lawyers’ rights and interests.

（四）    助力司法公开，让正义以看得见的方式实现

Facilitate Judicial Transparency and Let the Public Feel Justice

科技让公正在便捷中实现。上海高院坚持“公开是原则、不公开是例外”的原则，充分运用自然语言理解、机器学习、图文识别等人工智能新技术，打造了审判流程公开平台、裁判文书公开平台、执行信息公开平台、庭审网络直播平台、网络司法拍卖平台、司法监督平台等具有上海法院特色的十二大司法公开服务平台，建立了全方位、多层次、互动式、智能化的司法公开体系，构建开放、动态、透明、便民的阳光司法机制，以保障人民群众的知情权、参与权、表达权和监督权，让正义看得见、摸得着、可评价。目前，上海法院已向社会公开了112个方面830个信息项，累计发布信息2.35亿余条，日均访问量4.05万余人次。

Technology enables justice to be achieved in convenience. Guided by the principle of “opening to the public is the rule except particular case”, Shanghai high court make full use of new AI technologies, like natural language understanding, machine learning, image and character recognition identification, and has built 12 judicial service platforms with characteristics of Shanghai court, like trial process platform, judgment document platform, execution information platform, judicial trial live platform, online judicial auction platform and the judicial supervision platform and so on. Also, it has set up an all-around, multilevel and interactive and intelligent open judicial platform, and a open, dynamic, transparent and convenient sunshine judicial mechanism in order to guarantee people’s right to know, participate, express and supervise, making justice available, visible and evaluable. At present, Shanghai court has publicized 830 information items of 112 aspects to the society, and released a total of 23.5 million messages with more than 40,500 daily visits.

（五）    构建大数据分析平台，提升决策水平

Build a Big Data Analysis Platform and Improve Decision-making

大数据和“人工智能+”是科学决策的驱动力。上海高院建立了全国首个省级法院“新型司法智库”，依托司法大数据库，运用“人工智能+”技术，建立案件审判态势，民事、商事、金融、刑事、执行工作等多种审判大数据专项分析的智能分析平台，从海量的审判数据中分析发现审判规律，大大提升数据分析处理能力、知识发现能力和辅助决策能力，促进法院科学决策，提升管理水平，助力社会治理，平安上海、法治上海建设。

Big data and “artificial intelligence+” are the driving forces of scientific decision-making. Shanghai High People’s Court established the first provincial “new judicial think-tank”. Relying on judicial big database and “artificial intelligence+” technology, it established the intelligent analysis platform of trial elevation, civil, commercial, financial, criminal, execution and other trial big data. This platform can analyze and find trial law from the mass trial data, which can enhance the ability of data analysis and processing, knowledge discovery and auxiliary decision-making, so it can promote scientific decision-making of the court, improve its management level, facilitate social governance and construction of Shanghai in safety and the rule of law.

三、人工智能在推进以审判为中心的诉讼制度改革中的应用与实效

The Application and Result of Artificial Intelligence in the Reform of Litigation System Centered on Court Trial

（一）上海高院敢为人先，积极探索，承担研发“推进以审判为中心的诉讼制度改革软件”的任务

Shanghai High People’s Court Initiatively and Actively Explores and Undertakes the Task of “Promoting the Reform Software of Litigation System Centered on Court Trial”

推进以审判为中心的诉讼制度改革是党的十八届四中全会确定的重大改革任务。开发“推进以审判为中心的诉讼制度改革软件”是中央政法委交给上海高院的一项重要任务。

The reform of the litigation system centered on court trial is a major reform task set out in the Fourth Plenary Session of the 18th CPC Central Committee. It is also an important task from Central Political and Judiciary Commission to Shanghai High People’s Cuort.

2017年2月6日，孟建柱书记、韩正书记等在上海高院调研时，明确由上海高院承担研发“推进以审判为中心的诉讼制度改革软件”的任务（以下简称“206工程”）。这是一项把司法改革和现代科技融为一体，把大数据、人工智能融入刑事办案中的一项全新科技创新，在国内乃至全世界均没有案例可循，可见难度之大。

On February 6, 2017, when Meng Jianzhu, Secretary of Central Political and Judiciary Commission, and Han Zheng, Secretary of Shanghai Municipal Party Committee and so on investigated and surveyed in Shanghai High People’s Court, they determined that Shanghai High People’s Court undertakes the task of “Promoting the Reform Software of Litigation System Centered on Court Trial” (hereinafter referred to as the “206 Project”). This is an innovation that combines judicial reform and modern technology, and integrates big data and artificial intelligence into criminal case handling, which is rare at home and abroad, so it is rather difficult.

我们与上海公安、检察院密切配合，与科大讯飞公司合作，日以继夜、攻坚克难，圆满完成了研发任务。经过数月的努力（5月3日上线试运行），2017年7月10日，在全国司法体制改革推进会上，上海高院作了“上海刑事案件智能辅助办案系统”的汇报演示，得到了与会者的充分肯定和高度赞扬。孟建柱书记在讲话中指出，上海运用现代科技推进以审判为中心的刑事诉讼制度改革，具有“启示意义”，“已显现出广阔的发展前景”，“为第一个‘吃螃蟹’的，在荆棘丛生的改革路上找到了突破口，应为他们的创新鼓掌喝彩。”

We cooperated closely with the Shanghai Public Security Bureau, Shanghai People’s Procuratorate, and IFLYTEK Company, and successfully completed the research and development task. After months of hard work (trial operation on 3th May), July 10, 2017, at the promotion meeting of national judicial system reform, Shanghai High People’s Court showed a presentation of “Shanghai Intelligent Auxiliary System of Criminal Case Handling” which won participants’ full affirmation and praise. Secretary Meng Jianzhu in his speech pointed out that Shanghai’s practice has much significance. As “the first to try tomato”, it shows broad prospects. It has found breakthrough point on the thorny way of reform, and we should applaud for their innovation.”

（二）“上海刑事案件智能辅助办案系统”的技术原理

1、海量司法大数据为“上海刑事案件智能辅助办案系统”奠定数据源基础

随着互联网的飞速发展，在线数据变得异常丰富，多来源、实时、大量、多类型的数据可以从不同的角度对现实进行更为逼近真实的描述，为人工智能的落地应用奠定数据源基础，并通过大量数据构建人工智能的算法模型。[8]在推进“206工程”过程中，通过建立刑事案件大数据资源库，包括证据标准库、电子卷宗库、案例库、裁判文书库、法律法规司法解释库、办案业务文件库等子库，以及公检法三机关办理刑事案件的信息资源（办案）共享网络平台，能为人工智能应用提供强大的信息数据资源支撑和保障。

(B) The technical principles of “Intelligent auxiliary case handling system of Shanghai criminal cases”

1. Mass judicial big data lays the data source foundation for the “Intelligent auxiliary case handling system of Shanghai criminal cases”

With the rapid development of Internet, the online data has become extremely rich. The multi-sources, real-time, abundant and multi-types data can provide realistic approximations from different perspectives and lay the foundation for the landing application of artificial intelligence, and can build an artificial intelligence algorithm model through a large amount of data. ⁸In promoting the “206 Project”, it can supply artificial intelligence application with a powerful source of information and data resources support and guarantee by setting up a criminal cases’ big data repository, including the evidence standard repository, electronic file repository, case repository, judicative document repository, judicial interpretation repository of laws and regulations, case-handling business repository and other branch repositories, as well as information resource(case) sharing network platform of three public security authorities’ criminal cases handling.

2、深度学习算法是“上海刑事案件智能辅助办案系统”的核心引擎

机器学习尤其是深度学习，强化学习的完善和迭代促成了人工智能与各个领域的结合。在推进“206工程”过程中，关键要建立起完善刑事案件证据统一适用标准，通过人工智能深度学习算法，使软件具有逻辑思维和经验判断功能。运用图文识别（OCR）、自然语言理解（NLP）、智能语音识别、司法实体识别、实体关系分析、司法要素自动抽取等人工智能技术，通过对公、检、法已积累的刑事案件典型案例、司法信息资源、办案经验的深度挖掘、剖析提炼、机器学习，以确定的证据标准、证据规则和证据模型为基本遵循，实现对各种证据的智能识别、信息提取和逻辑关系校验等，以及时发现、及时提示进入系统的刑事案件中的证据标准不统一、办案程序不统一、证据中存在的瑕疵以及证据之间的矛盾等问题，实现防止冤假错案，减少司法任意性的目标。

2. Learning the algorithm in depth is the core engine of “Intelligent auxiliary case handling system of Shanghai criminal cases”

As to machine learning, especially the in-depth learning, improvement and iteration of intensive learning have led to the combination of AI and various fields. In the process of advancing “206 Project”, it is of crucial importance to establish a unified application standard of evidence in criminal cases and make the software have the function of logical thinking and experience judgment through the in-depth learning algorithm of AI. By applying AI techniques of Optical Character Recognition (OCR), Natural Language Processing (NLP), Intelligent Speech Recognition, Judicial Entity Identification, Entity Relationship Analysis and Automatic Extraction of Judicial Elements, and by deeply exploiting, analyzing and refining, and machinery learning of the accumulated typical criminal cases, judicial information resources and cases-handling experience of public security authorities, we achieve the intelligent identification, information extraction and logic checks of various evidences with assured evidence standard, evidence rules and evidence models as the basic follow-up, so as to promptly find out and remind you the inconsistent evidences and procedures, the existed defects in evidences and contradictions between evidences in the system’s criminal cases, thus false and unjust cases and arbitrary objectives can be avoided.

3、强大的计算能力使“上海刑事案件智能辅助办案系统”应用更加多元

伴随云计算技术和芯片处理能力的迅速发展，可以利用成千上万台机器进行并行计算，尤其是图形处理器（GPU）、可编程专用集成电路（FPGA）以及人工智能专用芯片（比如Google的TPU）的发展为人工智能落地奠定了基础计算能力，使得使用类似于人类的深层神经网络算法模型的人工智能应用成为现实。在推进“206工程”过程中，通过云计算、人工智能等技术，系统可以具备证据标准指引、单一证据审查、逮捕条件审查、社会危险性评估、证据链和全案证据审查判断、办案程序合法性审查监督、庭审示证、类案推送、量刑参考、文书生成、电子卷宗移送、全程录音录像、知识索引等多种功能，普遍提高办案人员水平。

3. Powerful computing competence makes “Intelligent auxiliary case handling system of Shanghai criminal cases” polynary

With the rapid development of cloud computing technology and chip processing capability, thousands of machines can be used for parallel computing, especially the development of Graphics Processing Unit (GPU), Field-Programmable Gate Array (FPGA) and AI specific chips (such as Google’s TPU) has laid the foundation of computing competence for AI’s landing so as to make the application of AI that is similar to the algorithm model of human’s deep neural network a reality. In the process of advancing “206 Project”, the system can be equipped with various functions, such as evidence standard guideline, single evidence censorship, arresting condition censorship, social risk assessment, evidence chain and case-based evidence censorship and judgment, censorship and supervision of case-handling procedure’s legality, court trial testimony, case propulsion, sentencing reference, document generation, electronic file transfer, video and audio recording and knowledge index by applying the technologies such as cloud computing and AI, and investigators’ ability can be generally promoted.

（三）“上海刑事案件智能辅助办案系统”取得的初步成效

1、建立了大数据资源库。截至2017年6月底，上海刑事案件大数据资源库已汇集1,695万条数据。其中，案例库中的案例9,012个，裁判文书库中的文书1,600万篇，法律法规司法解释库中的条文948,384条，办案业务文件库各类规范性文件（公检法）638件。证据标准库、电子卷宗库将随证据标准的制定及开发的案由同步更新。这些都为机器深度学习提供了丰富的数据资源。

2、制定了证据标准、证据规则指引。我们根据中央政法委的要求，结合上海司法实践，聚焦常见多发、重大、新类型等案件，特别是当前社会的公共安全以及人民群众迫切需要解决的问题，先行选择了7类18个具体罪名开展证据标准指引制定工作，[9]为办案人员提供了统一适用、简便易行、数据化、清单式的办案指引。同时，根据刑事诉讼法关于办案流程的不同要求，坚持完整性与阶段性相统一的原则，按照立案、逮捕、侦查终结、审查起诉、审判等不同办案阶段，分别制定了不同的证据标准指引，为公检法三机关在不同阶段提供了办案指引。

3、搭建了办案统一网络平台。在公检法三机关之间建立了统一的刑事办案平台，形成新的办案流程，消除了长期以来存在的“信息壁垒”，初步实现了刑事办案网上运行、互联互通、信息共享。

(C) Preliminary success achieved by “Intelligent auxiliary case handling system of Shanghai criminal cases”

1. Establishment of the big data repository. Till the end of June 2017, 16.95 million data has been collected in the big data repository of Shanghai criminal cases. Among them, there are 9,012 cases in the case repository, 16 million documents in the judicative repository, 948,384 articles in the judicial interpretation repository of laws and regulations, and 638 pieces of various kinds of normative documents (public security authorities) in the case handling business repository. Evidence standard repository and electronic file repository will be simultaneously updated with the development of evidence standards and cases. All of these provide abundant data resources for in-depth machine learning.
2. Development of evidence standard and evidence rule guideline. In accordance with the requirements of the Political and Judiciary Commission under the Central Committee of the Communist Party of China and combining Shanghai judicial practice, we focus on these frequent, significant and new-type cases, especially the ones related to public safety and people’s urgent needs. We first select 18 specific crimes under 7 categories to start the establishment of evidence standard guideline, ⁹providing investigators with a uniform, convenient, data-orienting and list-format case handling guideline. At the same time, according to different requirements of case handling procedure in the Law of Criminal Procedure, we adhere to the principle of unity between integrity and periodicity and establish different evidence standard guidelines respectively in accordance with the different phases of filing cases, arresting, investigation termination, censorship prosecution and judgment, to provide the three public security authorities with case handling guidelines in different phases.
3. Establishment of unified case handling network platform. We build a unified criminal cases’ handling platform among three public security authorities and form a new case handling procedure, so as to eliminate the long-term existed “information barrier” and preliminarily achieve online operation, inter-connection and information sharing of criminal case handling.
4. The success of trial operation

On May 3, 2017, “Intelligent auxiliary case handling system of Shanghai criminal cases” was formally put into trial operation (6 courts, 6 procuratorates, 13 public security departments, a total of 25 pilot organizations were launched). Since May 1, 4 crimes such as the crimes of intentional homicide, larceny, illegal absorbing of public deposits and swindling (telecommunication fraud) filed in these pilot organizations are all included in the system’s trial operation. By the end of October, the system has recorded 65 cases and 20192 evidences, offered 3361 evidence guidelines and 406 knowledge index searches and has found out 48 evidence flaws in all, with total clicks of 63000 times.

4、试运行见成效

2017年5月3日，“上海刑事案件智能辅助办案系统”正式上线试运行（法院6家、检察院6家、公安机关13家，共计25家试点单位上线）。试点单位于5月1日起立案的故意杀人罪、盗窃罪、非法吸收公众存款罪、诈骗罪（电信网络诈骗）4个罪名案件，均进入该系统试运行。截止10月底，系统共录入案件共计65件；录入证据20192份；提供证据指引3361次；发现证据瑕疵点48个；提供知识索引查询406次；总点击量达6.3万余次。

（四）“上海刑事案件智能辅助办案系统”开发应用的价值体现

1、软件的开发与应用，是推进以审判为中心的诉讼制度改革的重要内容和切入点。通过建立统一的证据标准、规则指引，发挥软件系统的校验、提示、把关、监督作用，可以更好地落实公检法三机关的办案职责，更好地体现我国“分工负责、互相配合、互相制约”的刑事诉讼原则。

2、软件的开发与应用，是现代科技在司法领域深度应用的重大突破。将大数据、人工智能等现代科技融入到刑事诉讼活动中，把统一适用的证据标准嵌入数据化的办案程序，是一项全新的科技创新任务，开启了人工智能在司法领域深度应用的先河，是0到1的突破。

3、软件的开发与应用，发挥了证据标准、证据规则指引、证据校验、审查判断等功能作用，确保提请逮捕、移送审查起诉的案件符合法律规定的标准，增强了办案人员的证据意识、程序意识、责任意识、人权意识，倒逼办案人员从案件一接手，就按照法律规定的证据标准和证据规则收集、固定证据，促进了办案质效的提升。

4、软件的开发与应用，克服了办案人员个人判断的差异性、局限性、主观性，提高了对证据审查判断的科学性、精准性、全面性，防止了“起点错、跟着错、错到底”，防止了司法的任意性，对于防范冤假错案的发生，确保无罪的人不受刑事追究，有罪的人受到公正惩罚，具有重大意义。

(D) The value of developing and applying the “Intelligent auxiliary case handling system of Shanghai criminal cases”

1. The development and application of software is the important content and pointcut of promoting the judgment-focused procedural system reform. By setting up a unified evidence standard and rule guideline as well as give a full play to software system’s functions of check, reminder, guard and supervision, so as to better implement the three public securities’ responsibility of case handling and better embody China’s criminal procedure principles “ Responsibility division, mutual coordination and mutual restraint”.
2. The development and application of software is a significant breakthrough of modern technology in judicial in-depth application. The integration of modern technologies such as big data and AI into criminal procedure and the embedding of unified evidence standards into digital case handling procedures is a new scientific and technological innovation task, which means the breakthrough of 0 to 1, commencing the history of AI’s in-depth application in the administration of justice.

3. The development and application of software gives full play to evidence standard, evidence rule guideline, evidence check and examination judgment, which ensures the validity of cases of arrest, transfer examination and prosecution, enhances investigators’ awareness of evidence, procedure, responsibility and human rights, and pushes investigators to comply with the evidence standards and rules under laws of collecting and fixing evidence from the very beginning, so as to improve the quality and efficiency of handling cases.
4. The development and application of software overcomes the problems of difference, limitation and subjectivity of investigator’s personal estimation and promotes the scientificity, accuracy and comprehensiveness of evidence’s examination and judgment, so as to avoid “starting wrong, following wrong and doing wrong” and avoid the arbitrariness of justice, thus it maintains great significance of preventing false and unjust cases, protecting innocent people from being criminally prosecuted and keeping guilty people receiving just punishment.

[1]麦肯锡全球研究院给出更加激进的预测，人工智能正在促进社会发生转变，这种转变比工业革命“发生的速度快10倍，规模大300倍，影响几乎大3000倍”。The McKinsey Global Institute shows more aggressive forecast: artificial intelligence is promoting social change which is “10 times faster, 300 times larger in scale, 3000 times larger in influence” than the industrial revolution.

[2]正如英国法学家J.奥斯汀所主张的“所谓‘法治’就是要求结论必须是大前提与小前提逻辑必然结果。”如果法官违反三段论推理的逻辑，就会破坏法治。载朱景文主编：《对西方法律传统的挑战》，中国检察出版社1996年2月版，第292页。

As British jurist John Austin put it, “the so-called ‘rule of law’ requires that the conclusion must be the logic result of major premise and minor premise.” If the judge violates the logic of syllogism, it undermines the rule of law. Zhu Jingwen, The Challenge to Western Legal Tradition, China Procuratorate Press, February 1996, P292.

[3]法律的自动售货机批判是指整个法律运作就如同一台“加工机床”，只要提供一定的材料，就一定会产生确定的产品。例如，德国法学家萨维尼就曾针对这种现象说，“罗马法学家的方法论具有一种除数学之外其他地方再不会有的确定性；可以毫不夸张地说，他们是用他们的概念来进行计算的。”

The vending machine critique of law means that the entire legal operation is like a “machine tool”. As long as the materials are provided, certain products will be produced. For example, German jurist Savigny has said that “the Roman jurists’ methodology has a certain certainty that is exclusive to mathematics. It is no exaggeration to say that they are using their concepts to calculate.”

[4]从人工智能司法应用的发展起步阶段看，人工智能专家正是选择三段论演绎推理进行模拟，由美国人沃尔特和伯恩哈德在20世纪70年代初开发了法律推理系统，[4]计算机以“如果A和B，那么C”的方式对三段论加以描述，使机器法律推理第一次从理论变为现实。

Judging from the early judicial application of artificial intelligence, artificial intelligence expert exactly choose syllogism to simulate the deductive reasoning. American Walter and Bernhard in the early 1970s developed a legal reasoning system where 4 computers based on “if A and B, then C” to describe syllogism, which for the first time made legal reasoning machine from theory to reality.

[5]（美）博登海默著：《法理学——法哲学及其方法》，邓正来、姬敬武译，华夏出版社1987年12月版，第478页。

Edgar Bodenheimer, Jurisprudence: The Philosophy and Method of the Law, translator: Deng Zhenglai and Ji Jingwu, Huaxia Publishing House, December 1987, P478.

[6]大规模知识库系统（KBS）开发就注意了思维结构的整合作用，许多具有内在联系的小规模子系统，在分别模拟法律推理要素功能（法律查询、法律解释、法律适用、法律评价、理由阐述）的基础上，又通过联想程序被有机联系起来，构成具有法律推理整体功能的系统。

When developing large Knowledge-base System (KBS), the integrated function of thinking structure was paid attention to. Many small-scale subsystems with inner link, on the basis of simulating legal reasoning function (legal query, legal interpretation, applicable law, legal evaluation, reason statement), are connected organically by association procedure, forming a whole system with function of legal reasoning.

[7]张保生：《人工智能法律系统的法理学思考》，载《法学评论（双月刊）》2001年第5期，第16页。

Zhang Baosheng, Jurisprudence of Artificial Intelligence Legal System, Law Review (bimonthly), No.5, 2001, P16.

[8] Just as Dr. Wang Jian, chairman of the Alibaba Group Technical Committee, said that the progress of AI comes from the continuous improvement of Internet infrastructure, while it makes no sense if AI is isolated from the Internet.

[9] So far, the evidence standard establishment of crimes of intentional homicide, intentional injury (causing death by intentional injury), robbery (murder during robbery), larceny, swindling (telecommunication fraud), illegal absorbing public deposits and fraud in financing has been completed, a total of 48 parts and 1039 evidence examination standards.

斯坦福法学院18届法学博士薛梦白协助编译