MATHEMATICAL AND NEURAL NETWORK MODELS OF CONTROL AND OPTIMIZATION OF ECONOMIC DEVELOPMENT

The article presents a concept for developing artificial intelligence (AI) based on a hybrid model that uses, on the one hand, verified mathematical models to calculate and forecast potential trend trajectories of long-term socio-economic development, and on the other hand, short- and medium-term models of crisis phenomena for training a neural network with its subsequent use to determine the real economic situation and develop an appropriate optimal policy for managing current socio-economic development. It is proposed to build AI on the basis of the Kolmogorov–Arnold neural network, which is increasingly used to create AI designed to solve physical and engineering problems, including those related to the management of various dynamic processes. The advantage of the hybrid model proposed for the first time is its complete transparency in relation to identifying cause-and-effect relationships between the main factors and output in the process of socio-economic development.