Modern society at the current stage of its development is characterized by the rapid development of technologies in various fields. Artificial intelligence, neural networks, virtual reality, robotics – all this describes the fourth industrial revolution, which radically changed the social, cultural, and economic environment. Humanity must be adapted to widespread changes of this kind, because automation and robotization of production can completely modernize the entire socio-economic environment in general and restructure the labor market in particular. Many market segments are already equipped with innovative technologies, especially the engineering and financial sectors, which cannot be said about the legal sphere.

In this paper, we have set a goal – to establish acceptable boundaries for the introduction of neural networks and automated systems for making legally significant decisions and technical and analytical support of professional legal activities. In view of this, we have set such tasks as: reviewing the practice of testing the introduction of LegalTech technologies at the present stage, assessing the prospects for their development, determining the boundaries of the permissible introduction of neural networks and other artificial intelligence technologies from ethical and humanistic positions. The set goals and objectives have justified the scientific novelty, which consists in the fact that such aspects remain poorly studied and insufficiently considered in modern doctrine, taking into account the intersection of various aspects of the development of law in the conditions of undulating and rapid transformations of scientific and technological progress.

Only in recent years, the LegalTech direction has been introduced into companies with a legal profile. One can observe a variety of legal search engines, design programs for drawing up standard contracts and lawsuits, chatbots that fully meet the basic needs of modern society, electronic document management, as well as the use of robots, artificial intelligence and neural networks. Today we are witnessing the equipping of only some organizations with such innovative technologies. Thus, Sberbank PJSC actively uses, develops and implements AI technologies that have gained the trust of specialists and customers[1]. In 2016, about 450 lawyers who were engaged in the preparation of claim documents were dismissed from the organization[2]. The neural network began to make claims with greater efficiency. The use of modern technologies finds its place in structural bodies, for example, the police use technologies for identifying and verifying persons using a neural network. This provides tremendous assistance in performing the function of searching for persons. Thus, a neural network is a human assistant in performing mechanical work that does not require creative approaches and solving unique tasks. The locality of its use is largely explained by the lack of legislative regulation of this area. It is considered necessary to create appropriate legal norms that would contribute to the integrated introduction of innovations into the justice system, which justifies the allocation of the improvement of the regulatory legal framework as one of the directions outlined in the NTI "neuronet" roadmap[3].

Artificial intelligence is the process of an automatic decision-making system based on an algorithm originally laid down. In this regard, we can assume that at this stage of its development, it will be able to cope with only part of the legal functions, namely, to perform typical tasks. In many ways, AI is able to do human work better, here are studies confirming this hypothesis. In the USA, highly qualified lawyers compared their skills with AI in interpreting legal documents. In addition to the fact that the latter did the work faster than specialists, he was able to analyze contracts related to damages, arbitration, confidentiality, more precisely lawyers by an average of 10%. In the UK, the best lawyers in London and the CaseCruncherAlpha chatbot, developed by three law students at Cambridge University, had to predetermine the decisions of the financial Ombudsman on 775 controversial cases concerning insurance payments. According to the percentage of the experiment result, artificial intelligence won by a margin of 20.3%. Consequently, research confirms that artificial intelligence is able to more accurately and quickly perform activities that require a certain algorithm, which means it does not involve a creative process. This can be proved by Sergey Pereverzev's project – "Legal Battle". A legal discussion was organized, within the framework of a simulated trial, by Roman Bezvenko with a chatbot based on neural networks[4]. The lawyer won the legal dispute. The result of such an event shows that lawyers face unique tasks every day that require an original approach and professional experience, neural network and artificial intelligence are highly specialized systems that currently do not fully cope with such functions. This leads to the question: so can all types of legal activity be replaced by AI?

Klaus Schwab believes that "professions such as lawyers, doctors ... can be partially or fully automated much earlier than one might assume"[5], and Roman Bevzenko is convinced that "in fifty years, robot lawyers, pumped with all the legal knowledge of the two-thousand-year history of jurisprudence, will completely displace people from 80 or 90 percent of positions: preparation of transactions, preparation of consultations, development of court documents"[6]. Undoubtedly, artificial intelligence will be used in legal activities, but in our opinion, it will be systemic only in the implementation of standard functions. Legal professions have specific tasks that require special psychological training, and personal contact of a specialist with people who need his help is important. Thus, a lawyer exerts a verbal influence on the jury in the trial, which can influence the final decision of the court. The mediator must have a set of personal qualities on which the success of the mediation process will depend, such character traits are: sociability, independence, organization, balance, concreteness, democracy, responsibility[7]. Judges carry out their activities guided by the principles of impartiality, objectivity, equality. Will artificial intelligence be neutral and objective in making court decisions? This is a controversial issue, because the AI system has its own autonomy, independence from humans, respectively, it is able to influence people, which can become a security threat to the whole world. The indicated specifics of the legal sphere make it clear that a person's personal qualities affect the legal proceedings, with which a neural network, artificial intelligence, robots will not be able to cope, this is the essence of the ethical-humanistic conflict of the introduction of AI into the legal sphere. In legal practice, each case is unique and requires a non-standard approach, which requires unconventional thinking and logical connections.

Thus, the introduction of automated systems into legal activity should have acceptable rational boundaries that allow leaving intelligent systems under human control. Undoubtedly, innovations can increase the efficiency of legal support of household and economic processes and even justice bodies in dealing with typical issues: debt collection, challenging fines, interpretation and analysis of regulatory legal acts and their projects, filling out documents, which will free specialists from routine work. In the next century, we will be able to observe the systematic and widespread use of artificial intelligence in the justice system, but completely replacing the legal profession does not seem acceptable due to the existence of irrational behavior of a person whose actions and inaction will come to the attention of specialized neural networks. This is due to the possible risks of introducing artificial intelligence, which has its own laws and language, which should become the basis for a harmonious combination of the capabilities of neural networks and humans.

List of used literature:

Decree of the Government of the Russian Federation No. 552-r dated March 30, 2018 "On approval of the Action Plan (Roadmap) for Improving Legislation and removing Administrative barriers in order to ensure the implementation of the National Technological Initiative in the direction of "Neuronet"" // URL: https://www . garant.ru/products/ipo/prime/doc/71818300 /
URL: https://www.pgplaw.ru/analytics-and-brochures/articles/roman-bevzenko-how-i-beat-trial-of-robot-lawyer-and-why-it-will-not-last-long/?ysclid=lb1bj2y4wr390905162
URL: https://stimul.online/news/v-sberbanke-rossii-yuristov-zamenili-na-neyronnuyu-set/
URL: https://www.garant.ru/products/ipo/prime/doc/71818300/
URL: https://www.pgplaw.ru/analytics-and-brochures/articles/roman-bevzenko-how-i-beat-trial-of-robot-lawyer-and-why-it-will-not-last long/?ysclid=lb1bj2y4wr390905162
URL: https://www.sberbank.com/ru/sustainability/principles-of-artifi cial -intelligence-ethics
Life 3.0. Being a human being in the era of artificial intelligence / Max Tegmark; translated from the English by D. Bayuk. – M.: "AST: CORPUS", 2019. – 560 p.
Korneeva Ya. A., Rudakova Yu. N. Personality features of mediators with different success of the conciliation procedure // Psychology and psychotechnics. 2019. No. 3. p. 68. .
The fourth Industrial Revolution / Shvab K. – M.: "Eksmo", 2018. – 138 p.