NEURAL NETWORKS: APPLICATION, ETHICS AND LAW ISSUES

Abstract

The use of artificial neural networks in practice has grown rapidly in recent years; they
are increasingly used in business, medicine, and even government administration. This proliferation
of neural networks as a decision-making tool changes the content of a number of social relations,
thereby posing new tasks for lawyers regarding the creation of legal regulation adequate to the
changes taking place. Based on the analysis of theoretical provisions and practical features related to
the training of artificial neural networks for the purpose of their further use as a tool for finding optimal solutions, the article draws conclusions about key issues of an ethical and legal nature, without
the resolution of which the negative effects of the use of neural networks will manifest themselves
significantly stronger.

One of these issues is the need for “fairness” in training an artificial neural network so that it
does not reproduce social prejudices and preferences of developers, discriminating against certain
social groups. Another question concerns the transparency of decision-making by a neural network;
the lack of understanding of the logic in choosing a solution does not contribute to establishing trust
on the part of people. No less important is the question of maintaining data confidentiality: how to
protect personal data if the neural network needs it to make a decision? The results of the study allow
us to more clearly formulate issues that require the priority attention of legal scholars in order to
search for solutions to them by the legal community. After all, the use of neural networks by practicing lawyers in their professional activities within the LegalTech direction, in parallel with the study
of theoretical aspects, allows them to get closer to the optimal solution to the issues raised.