Искусственный интеллект в отправлении правосудия: риск или будущее судебной системы?

Emelyanova M.A.,
Student of Russian State University of Justicenamed after V.M. Lebedev

Scientific consultantSokolova A.A.,
Senior Teacher at the Foreign Languages Department,
Russian State University of Justicenamed after V.M. Lebedev

Artificial intelligence in the administration of justice: risk or the future of the judiciary?

Abstract. The main purpose of this paper is to ascertain whether artificial intelligence technologies can be involved in the process of administering justice and decide on verdicts. The article analyses the specificity and features of artificial intelligence that prevent its application as a substitute for a judge. The foreign experience of the usage of the software that allows predicting the risk of recidivism is also considered. As a result, it was established that artificial intelligence is incapable of effectively performing the functions of the administration of justice and should be applied as an auxiliary algorithm.

Keywords: justice, artificial intelligence, judge, personalised approach, recidivism, inner conviction, legal consciousness, law enforcement.

In the modern world, there is a widespread opinion that artificial intelligence is a so-called panacea and, moreover, in the near future, it will be able to replace humans in many relevant spheres of activity. Due to the fact that jurisprudence is a significant sphere of social life, the question of whether artificial intelligence can be used in the process of administering justice and even deliver verdicts in the role of a judge is particularly substantial.

Justice is an activity providing the highest level of protection of violated rights and interests that is intended to resolve conflicts that have appeared in society. As the famous theologian and philosopher Aurelius Augustine stated, “where there is no justice, there is no civilization” [2]. Yet, can artificial intelligence, without possessing human traits, determine the fate of people?

Firstly, in order to understand the peculiarities of the functioning of artificial intelligence, it is necessary to formulate a definition of this concept. Artificial intelligence is a technology that allows computers and machines to imitate human learning, understanding, problem-solving, decision-making, creativity, and autonomy [7]. Artificial intelligence is based on a mathematical model that is specifically trained on a set of data to recognise certain patterns and use them to complete suggested tasks. Consequently, it cannot be unequivocally stated that the answer given by artificial intelligence is the correct one. Such networks are imperfect and lack critical thinking. In judicial procedure, each dispute requires an individual approach, and “fitting” cases to standard rules is impossible, and moreover, unethical. A striking example is the COMPAS (Correctional Offender Management Profiling for Alternative Sanctions) software used in the USА, which predicts the risk of a prisoner committing a criminal offence within two years of release. A COMPAS score of 1 to 10 assesses the risk of reoffending as low (1 to 4), medium (5 to 7), or high (8 to 10), and judges have repeatedly used these scores when sentencing. However, human rights activists who have studied the practice of applying this programme were doubtful about its impartiality. COMPAS errors have had a significant impact on black defendants: those individuals who did not reoffend were incorrectly predicted to reoffend at a rate of 44.9%, almost twice as likely as their white counterparts (23.5%); whereas white defendants who did reoffend were incorrectly predicted not to reoffend at a rate of 47.7%, almost twice as likely as their black counterparts (28%). In other words, COMPAS scores appeared to favour white defendants over black defendants, underestimating the likelihood of recidivism for white people and overestimating the likelihood of recidivism for black defendants [1]. Dartmouth College researchers decided to test whether predictions using COMPAS are more accurate than predictions of humans. They conducted an online survey whose participants were not involved in criminal justice. With significantly less information than COMPAS owns (just 7 traits compared to COMPAS’s 137), a small group of people predicted recidivism as accurately as COMPAS: the average accuracy of the people’s predictions was 62.1%, which is compatible with COMPAS’s prediction accuracy of 65% [4]. It is impossible to deny that the usage of such a technology is a major step in the development of digital justice, but it still has its disadvantages and requires development. For this reason, judges should not rely entirely on COMPAS, but use it as a supplementary tool to determine the chance of recidivism.

Another issue with the use of artificial intelligence is that, at this stage, it cannot compare to human thinking, as it performs tasks only according to logic and data without emotional context. A judge decides on a verdict on the basis of inner conviction, which is influenced by conscience and morality, because this is the only way a judge will be able to correctly correlate the evidence and the requirements of morality [5]. Moreover, the administration of justice must be guided by the principle of humanism, which consists of protecting the human being as the supreme value and minimising the repressive element: justice is designed to restore violated rights not only from a formal point of view. However, if we turn to the essence of artificial intelligence itself, we can conclude that a trial with its participation may become a mere mechanical comparison of facts with the “letter of the law” without understanding the “spirit of the law”, the law itself, which is considerably broader than the rule of law [6]. For the reason that artificial intelligence is not endowed with legal consciousness and has no emotional component, it is used merely as a “consultant” for the judge. For example, China has been working on a “smart court” system since 2016 after Zhou Qiang, the president of China’s Supreme People’s Court, called for the use of technology to improve the “fairness, efficiency, and authority” of courts. The system, which uses machine learning, automatically scans court cases for references, recommends laws and rulings to the judge, drafts legal documents, and corrects inaccuracies in them [8]. Nevertheless, it is the judge who performs the main role in decision-making. This shows that at the moment artificial intelligence is not able to replace the human judge, no matter how it may seem at first glance.

In conclusion, artificial intelligence is certainly capable of improving the efficiency of the judicial system, but its use instead of a judge seems impossible and even dangerous: judges do not simply apply the law, they interpret it in the context of complex life situations, taking into consideration human emotions, motives and social consequences of their verdicts. Artificial intelligence lacking these qualities runs the risk of delivering formal but unjust or unethical verdicts. Justice, on the other hand, is not mere application of algorithms but an art that requires human understanding and compassion.

References

1. Angwin, J., Larson, J., Mattu, S., Kirchner, L. Machine bias: There’s software used across the country to predict future criminals. And it's biased against blacks / ProPublica. URL: www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing (accessed 16.02.2025).

2. Augustine Aurelius. On the hail of God: translation from Latin. Moscow: AST, 2023.

3. Data-driven crime prediction fails to erase human bias. URL: https://www.sciencenews.org/blog/science-the-public/data-driven-crime-prediction-fails-erase-human-bias (accessed 16.02.2025).

4. Dressel, Julia, Farid, Hany. The accuracy, fairness, and limits of predicting recidivism / URL: https://www.science.org/doi/10.1126/sciadv.aao5580 (accessed 16.02.2025).

5. Markov, K.I. Moral aspect of the formation of the judge's inner conviction as a necessary component of judicial discretion in the process of verdict determination / K.I. Markov // Glossa: Bulletin of student science. Publication of the Department of Theory and History of State and Law of Kursk State University. 2023. № 7.

6. Ryabtseva, E.V. The problem of the use of artificial intelligence in criminal justice // Russian criminological journal. 2023. №1. URL: https://cyberleninka.ru/article/n/problema-ispolzovaniya-iskusstvennogo-intellekta-v-ugolovnom-pravosudii (accessed 17.02.2025).

7. What is artificial intelligence (AI). URL: https://www.ibm.com/think/topics/artificial-intelligence (accessed 16.02.2025).

8. Zou, C. X. Achievements and Prospects of Artificial Intelligence Judicature in China / Chinese Studies, 2020 / URL: https://www.researchgate.net/publication/363863187 (accessed 17.02.2025).