Artificial neural networks in economics: mathematical tool, model or methodology?

The purpose of this article is to assess the current interaction between artificial intelligence (AI) and economic science and to identify promising interdisciplinary areas of research that can significantly influence the methodology of understanding economic phenomena. To achieve this goal, the vague and partly even mystical term AI was replaced with а more scientific term «artificial neural networks» (ANN). The article uses methods of scientometric, epistemological and comparative analysis of the processes of ANN penetration into economics and other academic disciplines. The authors reveal the epistemological commonality and difference between AI and ANN and justify the shift in epistemological focus in research from general AI to ANN. The paper systematizes the use of ANN in economics: 1) ANN as a mathematical tool for solving economic problems; 2) ANN as a model of economic phenomena; 3) ANN as a methodology for understanding economic patterns. It shows the interaction of economics with neurosciences which occurs in two significantly different directions: from neurobiology, i.e. real nerve networks in living organisms, and, on the other hand, from ANN theory. The first direction is associated with neuroeconomics, the second has not yet been articulated, but shows an exponential growth in publications and is associated primarily with forming a new economic paradigm. The ANN paradigm in economics (and not only in economics) changes both the subject of cognition, introducing radically new forms/types of evidence and new research methods, and the object of cognition, changing the focus of study from individual economic behavior to the collective economic behavior of mega-subjects.

1. Andreyuk, D. S., (2012). Behavioral programs and information flows: a neuroevolutionary approach to optimizing the management of economic systems. Neurocomputing paradigm and society. Lomonosov Moscow State University.

2. Bakhtizin, A., Bragin, A., & Makarov, V. (2023). Large Fourth-Generation Language Models as a New Tool in Scientific Research. Artificial societies, 18(1). DOI: 10.18254/S207751800025046-9

3. Egorov, D. (2021). How Many Paradigms There are in Economic Theory — and Can It Become a One-Paradigm Science. Obshchestvennye nauki i sovremennost’, 5, 129–142. DOI: 10.31857/S086904990017287-7

4. Glazyev, S. Y. (2016). A New Paradigm of Economic Science. Public Administration. E-journal (Russia), 56.

5. Gubarev, R. V., Cherednichenko, L. G., Borodin, A. I., & Dziuba, E. I. (2023). Comparative analysis of the effectiveness of correlation-regression and neural network modeling in predicting energy emissions of carbon dioxide in Russia. Lomonosov Economics Journal, 58(3), 217–238. https://doi.org/10.55959/MSU0130-0105-6-58-3-11

6. Kovalevskaya, М. S. (2019). Scientific status of the economic model as an actual methodological problem of economic science. Ekonomika ustojchivogo razvitiya, 1(37), 349–353.

7. Neurocomputing paradigm and society. (2012). Ed. by Yu. Petrunin. Moscow: Moscow University Press, 304 p. (Scientific Research).

8. Peskova, A. V., & Kovalevskaya, M. S. (2016). Neuroeconomics and behavioral economics: synthesis sources. Bulletin of the South Ural State University. Ser. Economics and Management, 10(3), 18–25.

9. Polanyi, M. (1958). Personal knowledge. Towards a Post-Critical Philosophy. Chicago: University of Chicago Press. Edited by Mary Jo Nye.

10. Shastitko, A. E., & Tutov, L. A. (2023). Notes on methodological individualism in economic research: are there limits to the application? Lomonosov Economics Journal, 58(1), 3–21. https://doi.org/10.55959/MSU0130-0105-6-58-1-1