ETHICAL FOUNDATIONS OF ARTIFICIAL INTELLIGENCE TRAINING FOR THE RESOLUTION OF LEGAL DISPUTES IN THE RUSSIAN FEDERATION: OBJECTIVE LIMITATIONS FOR THE APPLICATION
Abstract
The article is devoted to the study of the ethical foundations of artificial intelligence (AI) training for the resolution of legal disputes in the Russian Federation (RF). This area is becoming particularly important in the context of modern digital transformation. The author formulates the author’s definition of this concept as a set of principles, norms and values governing the formation of intellectual systems capable of performing legal functions while maintaining a balance between technical efficiency and respect for legality, impartiality and responsibility to humans. Based on the analysis of current legislation, it is concluded that there is no direct legalization of AI in legal proceedings, which excludes the delegation of authority to administer justice to neural networks. The paper systematizes six ethical principles of AI learning: transparency and explainability, fairness, safety, responsibility, privacy, and respect for human rights. Four groups of objective limitations of the use of AI in legal disputes have been identified: technical (cognitive opacity, dependence on data quality), legal (uncertainty of the status of AI decisions and the subject of responsibility), ethical (risks of algorithmic bias, threat to transparency of justice) and social (institutional distrust, psychological barrier, risk of job cuts). Special attention is paid to the analysis of the «black box» problem and the need to preserve human responsibility for justice. The results of the study show that AI should perform exclusively auxiliary functions, while ultimate responsibility and morality remain the prerogative of humans. The proposed conclusions can be used to improve the regulatory framework, develop quality standards for training data, and enhance the digital competence of the legal community.