LARGE LANGUAGE MODELS USE IN THE CONTEXT OF RULEMAKING DIGITALIZATION

The main purpose of the research is to test the possibility of using generative language models (LLM-type neural networks) in rule-making activities, in particular, testing russian and foreign service for generating acts of rulemaking. The methodological basis consists of general scientific and special legal methods, including legal modeling, legal forecasting, extrapolations, as well as tools of the anthropological concept and convergent technologies. The scientific work considers fundamental issues related to the use of LLM in rule-making, such as: the problem of heterogeneity of the legal status of generative language models, approaches to digitalization of rulemaking in the context of virtualization of legal personality. The main result of the study is the concept of a digital person as a theoretical, legal basis for the development of mechanisms for the legal regulation of the use of LLM in rulemaking activities, as well as a list of risks arising during the experiment on the digitalization of rule-making based on self-learning algorithms: legal, ethical, communicative, methodological.