Сomparative legal study of artificial intelligence positioning in judicial proceedings.

Morhat Petr Mechislavovich Doctor of Law Judge; Arbitration Court of the Moscow District 107053, Russia, Moscow, Prospekt Akademika Sakharova str., 18, room 604 judiciary2024@yandex.ru Other publications by this author

Over the past decade, artificial intelligence (AI) has been actively integrated into the sphere of public relations in various countries, strategic documents are being formed in the field of regulating the use of AI in various sectors of the economy and law^[5]. In this logic, a trend is being formed and developing towards the implementation of justice using intelligent assistants, the development of a comprehensive regulatory approach to the safe implementation of this technology in the judicial sphere.

Previously, we investigated the phenomenon of artificial intelligence (hereinafter - AI) in the context of the possible involvement of AI units in the judicial process^[1]. Considering the role and place of AI in the judicial process, we noted the possible participation of this technology at the stage of prudential control, information and documentary support of legal proceedings, provision of expert-analytical intellectual legal proceedings, linguistic and other analysis^[2].

For this work, the research methodology is based on systemic, structural-functional, hermeneutical and comparative legal methods. Within the framework of the systemic and structurally functional method, the positions of states and law are studied in the context of the application of AI in the field of judicial proceedings, the function of law and legal responsibility, through the prism of decision-making by a judge using AI technology. The hermeneutical method is reflected in the analysis and interpretation of the positions of the courts regarding the admissibility of the use of AI technology in the judicial process. The comparative legal method (the method of legal comparative studies) is the main method within the framework of this article, since the analysis of current approaches related to the regulation of AI in foreign countries (China, USA, France) and Russia is actively used.

Considering the technology of artificial intelligence, it should be noted that this technology penetrates deeper into the sphere of public relations and arouses the corresponding interest both among researchers from a regional perspective^{[3, 4]} and at the level of international organizations^{[5, 6]}. In this regard, it is proposed to investigate the AI phenomenon in more detail in the context of the judicial process of the world's leading countries.

At the level of the leading countries, it is impossible not to mention the experience of China. In 2017, China published its Artificial Intelligence Development Plan, which outlines China's strategic goal setting until 2030 in the field of AI technology development. It should be noted that within the framework of long-term AI planning, the use of AI in new forms of "electronic management" is envisaged for the horizon of 2030^[7].

Considering China's experience in detail, it should be noted that the country's national plan on the regulation of AI technology provides a brief paragraph on the possibilities for future "smart courts", indicates the creation of an application (or a set of IT solutions) to ensure litigation, interactive support of the trial, including evidence collection, case analysis, reading and analysis legal documents.

Obviously, providing an appropriate database for AI training requires the formation of a "data bank".

In 2014, the Supreme People's Court of China declared "openness" and "judicial transparency" as key components of the judicial process for this purpose^[8]. The government platform "China Judgments Online" currently hosts more than 130 million public documents^[9]. This data bank contains additional sub-platforms "Similar Court Decisions" and "China's Big Data Justice Service Platform", which allow judicial officials and members of the public to search for data on court decisions^[10].

In the future, it is necessary to note the formation of the revolutionary AI system "System 206" in Shanghai, which has cognitive functions based on machine learning^[11]. At the stage of filing documents, this system checks the criteria for accepting the case, an intellectual reminder of the payment of the court fee and the intellectual formation of procedural documents; at the stage of pre-trial analysis, it identifies the essence of claims, determines the position of the defense, identifies the facts and causes of the dispute, analyzes the case materials, verifies the completeness and reliability of evidence and prepares a plan for holding a court session with a preliminary at the stage of the trial, establishes indisputable facts, the positions of the parties, conducts paperless cross-examination and prepares the minutes of the court session; at the stage of making a decision, AI helps in selecting the legislative framework, similar court cases, prepares a preliminary decision prepared by analogy with another case, as well as related documents (for which separate subsystems have been created).

As part of the pilot operation of the AI, 495864 cases were processed in 2016 alone, by April 2019, the AI was processing and detailing cases related to road traffic disputes, credit cards, equity participation, dissemination of information on the Internet, etc. This tool significantly simplified the work of judges and indirectly influenced the reduction of the judicial corps in 2017, the number of judge quotas from 210,000 to 120,000 people^[12].

As of 2021, the automatic speech recognition technology underlying the 206 system provided by iFlytek has been implemented in more than 4,200 courtrooms across the country^[13].

In addition to the above-mentioned AI decisions, more attention needs to be paid to prospective AI judges who directly assist in making court decisions (moving from assistants to potential participants). In Hangzhou, the AI judge "Xiao Zhi" was used to adjudicate a private loan dispute, helping a human judge review the case and make a decision in less than 30 minutes^[14]. This AI helps judges in real time with an interactive pool of arguments, evaluation of evidence and recommendations for a decision.

These AI decisions, for all their merits and decision-making speed, are carried out under the close supervision of human judges, and no court decision is executed without human approval.

The position of Zhou Qiang, Deputy Chairman of the 14th convocation of the CPC Central Committee (previously Chairman of the Supreme People's Court of China from 2013 to 2023), is that "AI will never replace human judges and can only serve judges (as assistants)." At the same time, Zhou Qiang calls for more intensive use of AI technology in the judicial process, to consult with AI when considering all categories of cases^[15]. If the judges reject the recommendations of the AI, they must provide a written explanation of their actions^[16].

With all these advantages, it is impossible not to note certain disadvantages of AI technologies for the purposes of the judicial process. For example, current pilot programs testing facial and emotion recognition technology to confirm the reliability of readings are unreliable and technologically imperfect at the moment^[13]. There are also issues of behavioral adaptation (which is formed in potential offenders) and uniformity of judicial practice, which is a source of processing for AI. In this situation, we understand that the AI analyzes the situation based on a certain model formed in the array it previously processed. Bearing in mind the indicated mandatory practice of providing explanations from judges about the reason for disagreement with the recommendations presented by AI, as well as taking into account the function of algorithmic monitoring of general judicial practice and deviations from it implemented by AI, we can talk about the risk of a certain bias on the part of judges in "cases falling out of the general statistical mass" and the difficulty of making decisions solutions that contradict the recommendations of AI^[17].

At the same time, in our opinion, this approach is more constructive than completely blocking the development of AI, including from the perspective of predictive litigation in France^[18].

Considering the experience of the United States, it should be noted that the legislative framework is defined at the level of the National Artificial Intelligence Initiative 2020, according to the provisions of which the United States should ensure leadership in the field of AI, as well as maintaining the leading positions of the United States in the world in the development and use of reliable artificial intelligence systems in the public and private sectors^[19].

Considering the use of AI in the judicial process, it is necessary to note the experience associated with the COMPAS (Correctional Defender Management Profiling for Alternative Sanctions) system, which is actively used to assess the risk of recidivism in convicts applying for parole. This system, based on a large number of parameters, including gender, age, education, kinship and social environment, the article on which the defendant was convicted, the number of times prosecuted, etc. (more than 130 factors in the database), allows us to determine with a high degree of accuracy the potential risk of the defendant to society in the event of a positive decision by the judge.

In this regard, the case of the State of Wisconsin against Eric L. Loomis, where the Computer c program was used, is indicative^[20]. According to the court decision, when assessing the identity of Eric L. The AI issued a negative opinion against the defendant. Within the framework of the process itself, a noteworthy factor is that in the future, within the framework of the appeal procedure, the legality of the use of this program was not disputed, but the incorrectness of its algorithms in relation to the assessment of the defendants was emphasized.

Considering the practice of using AI in the judicial process, it is impossible not to note the possibilities of modern AI technology for the preparation of statements of claim. In the case of Mescall v. Renaissance at Antiquity Slip Copy, the plaintiff's lawyer used ChatGPT to write his lawsuits^[21]. In this particular lawsuit, the court ruled that "the use of artificial intelligence creates problems, raises ethical concerns and can lead to sanctions or fines."

In its conclusion, the court also referred to the practice of Mata v. Avianca, Inc. where it is noted that the legal adviser, on the basis of a conscious action, misleads the court when he provides non-existent judicial opinions with fake citations and links created by the ChatGPT artificial intelligence tool^[22]. These actions are grounds for imposing appropriate sanctions against the party by the court.

In the case of Williams v. the City of Detroit, a U.S. citizen from Michigan in 2020 became a victim of such bias on the part of AI technology used in law enforcement^[23].

The case began with Mr. Williams interpreting the call from the police department as a joke. The call from the police department is due to the fact that his identity was mistakenly identified using a facial recognition system, and he was charged with a crime he did not commit.

When the police detained Mr. Williams, they showed him a photo of the surveillance footage in question, and Mr. Williams commented to the police department that all black men looked the same in the "grainy" photos. Later it turned out that the facial recognition system was wrong and the police arrested the wrong man.

Considering the case of the use of AI in the framework of algorithmic analysis of personal data, it is necessary to refer to the case of Tice v. Amazon.com Inc. The plaintiff filed a lawsuit against Amazon, claiming that Alexa (a virtual assistant developed by Amazon and used in Amazon Echo and Amazon Echo Dot smart speakers) violated the California Invasion of Privacy Act by recording private conversations without the plaintiff's knowledge^[24].

In this case, the court ruled that the recording by the AI was limited when they used the column, and there was no confidentiality agreement between Amazon and the plaintiff, according to which Amazon assumed responsibility for "damages" related to the use of the column. Thus, the court was unable to satisfy the plaintiff's claims due to the blurred line related to the user agreement and the privacy law. A similar practice was presented in Wilkoski's class action lawsuit against Amazon.com Inc. (illegal recording of speaker users' conversations), in connection with a violation of the Illinois Biometric Privacy Act^[25].

Considering the phenomenon of artificial intelligence in the judicial practice of Russia, it is impossible not to note the strengthening of the regulation of its use for public administration purposes.

As part of his latest address to the Federal Assembly, President of the Russian Federation Vladimir Putin notes the significant role of AI technology for building optimal interaction between citizens, business and the state (Message from the President to the Federal Assembly. URL: http://kremlin.ru/events/president/news/73585 (date of application: 07/10/2024). This presentation also noted the updating of the strategic document in the field of AI regulation and the establishment of a regulatory horizon until 2030 (Decree of the President of the Russian Federation dated October 10, 2019 No. 490 "On the development of artificial intelligence in the Russian Federation" (with amendments and additions).

The strategy stipulates that the main task of using AI is to automate the same type and repetitive production operations, including operations carried out in the field of administration of justice. Within the framework of this case, the starting point for the purposes of the trial is the Online Justice service being created, which allows you to submit documents to the court in electronic form, study the course of the trial and participate in it online, as well as receive electronic court documents signed with an electronic digital signature of the judge^[26].

In addition, participation in the court session of the parties is possible using web conference technology due to the introduction of biometric authentication technology for participants in the process by face and voice. Within the framework of this service, automated creation of draft court decisions is expected based on the analysis of the text of the procedural document and case materials.

As part of the current legislation, since September 1, 2019, automated distribution of cases has been introduced in Russia (Article 18 of the APC of the Russian Federation, Article 14 of the CPC of the Russian Federation, Article 28 of the APC of the Russian Federation, Article 30 of the CPC of the Russian Federation). The system operates both in arbitration courts and in courts of general jurisdiction for a fair and impartial distribution of workload among judges.

Member of the Presidium of the Supreme Court of the Russian Federation V.V. Momotov notes that "weak AI" is already actively used in the court system. AI helps to draft court decisions, verifies the completeness of the package of documents submitted in the case. And as part of the experimental application of AI in the Amur region, the plaintiff can formulate his statement using the appropriate technology^[27]. At the same time, it is noted that a strong AI is unacceptable as a participant in the judicial process and only a judge has such necessary qualities as justice, respect for participants, decency, dignity, and respect for professional secrecy.

In the same context, it is necessary to note the position of the Chairman of the Constitutional Court of the Russian Federation V.D. Zorkin, who emphasizes that AI is able to implement actions leading to legally significant results and act as a bridge for the partial transition of justice to the online sphere. At the same time, the danger of the construction associated with granting AI personal rights by analogy with human rights is noted ^[28].

The author of the article supports the approach of V.D. Zorkin on the inadmissibility of equalizing AI technology in the status of a legal entity identical to the status of a person and a judge. At the same time, it should be noted that this technology, due to its potential autonomy and the formation of legally significant consequences, will require the development of fundamentally new approaches, including through the use of existing mechanisms or the formation of new constructs. Otherwise, in our opinion, a situation will be formed when the AI developer (or another person) intends to underestimate the autonomy of the specified technology in order to offset the risks of liability or complicate the procedure for establishing legally significant facts to determine the responsibility of individuals.

In this logic, a possible solution tool is the extension of the current structure of legal entities to AI^[29] or the development of the concept of an electronic person for the purposes of public law^[30].

The author supports the thesis presented in the article by G.A. Gadzhiev and E.A. Voynikanis that in modern legal science there is already a need to resolve legal disputes involving autonomous robots (robots with AI), and the possible technique of legal assimilation of AI to legal entities (as the most elastic structure) or quasi-legal entities is one of the ways effective consideration of these disputes ^[29].

An alternative approach to the development of the concept of an electronic person for the purposes of public law, presented by O.A. Yastrebov and M.A. Aksenova, is based on a legal fiction in the form of a certain organizational structure using AI, which is proposed to be fixed as an electronic person with subsequent classification for the purposes of public law. At the same time, the controversial nature associated with the definition of legal responsibility and the conditions for granting legal personality to an electronic person is also noted ^[30].

Based on the above, the following conclusions and suggestions can be formulated:

1. The introduction of artificial intelligence into the domestic judicial process is a prospect for the near future, which will require the formation of an appropriate legislative framework and judicial practice that takes into account the features of this technology in all technical aspects, addressing issues of reliability of the proposed recommendations and solutions of AI, as well as ensuring appropriate information security.

2. Possible areas of AI implementation within the framework of domestic judicial practice differ little from global trends, however, issues of assessing the evidence base from the AI side will require additional independent external expertise or the formation of a public institution that monitors AI technologies used in the field of public authority.

3. The starting point for the most effective and safe implementation of AI can be cases related to a simplified procedure for considering a case or those disputes that are of a high nature of reliable digital data.

4. Speaking about the possible legal personality of AI technologies for the purposes of the judicial process, at the moment it is necessary to note the inadmissibility of using a design identical to a person. However, a possible alternative solution may be a legal fiction through the prism of a legal entity or the creation of a new design of an electronic person.