Neural network as a mirror of social attitudes: analysis of distortions in generative images

The article is devoted to the consideration of neural network generative technologies as a marker of social stereotypes and attitudes. The aim of the research – approbation of generative artificial intelligence (hereinafter referred to as AI) as a method of sociological research of social stereotypes contained in big data. To realise this goal, the essence of AI, the legal framework of application and the spread to date are initially considered. The results of approbation show that the information returned by AI contains social stereotypes, primarily related to gender and age, which means that AI can indeed be used as a tool for studying social stereotypes. The source of shifts in data towards stereotypical images is contained in the data on which AI is trained, as well as in the code of the program itself, that is in the attitudes and worldview of developers, which in one way or another influence the process of program development. In most cases (more than 80% of all generated information), the AI returns young people, predominantly men, for queries related to high-paying professions, which is true for both gendered and non-gendered query formulations. AI is also characterised by attributing certain traits to different social groups, such as slovenliness and disorganisation, representing them in connection with a certain style of dress, and using several recurring markers to denote status or wealth.

1. Bashirova L.S. Social norm and deviation. Psychopedagogy in law enforcement agencies. 2000;2(14):97–100. (In Russian).

2. Chviakin V.A., Grigoriev N.Yu., Konoplin Yu.S. Cognitive sense of social stereotypes. Humanitarian of the South of Russia. 2023;4(12):94–103. (In Russian). https://doi.org/10.18522/2227-8656.2023.4.5

3. Lim W.M., Gunasekara A., Pallant J.L., Pallant J.I., Pechenkina E. Generative AI and the future of education: Ragnarok or reformation? A paradoxical perspective from management educators. The International Journal of Management Education. 2023;2(21). http://dx.doi.org/10.1016/j.ijme.2023.100790

4. Lippman U. Public opinion. Moscow: Institute of the Public Opinion Foundation; 2004. 382 р. (In Russian).

5. Malyshev I.O., Smirnov A.A. Review of modern generative neural networks: domestic and foreign practice. International Journal of Humanities and Natural Sciences. 2024;1–2(88):168–171. (In Russian). http://doi.org/10.24412/2500-1000-2024-1-2-168-171

6. Melnikova D.A., Lopatkin D.S., Kozheva A.A. Artificial intelligence as a way to create new content. Advances in chemistry and chemical technology. 2023;1(263(37):43–47. (In Russian).

7. Piterova A.Yu., Teterina E.A. Social stereotypes: features of formation and study. Science. Society. State. 2016;1(13). (In Russian).

8. Popkov V.D. Stereotypes and prejudices: their influence on the process of intercultural communication. Journal of Sociology and Social Anthropology. 2002;3(5):178–191. (In Russian).

9. Smirnova Y.S. Relaxation of prejudices and development of tolerance as a problem of formation of professionally significant qualities of the future specialist. In: Principle of tolerance and its application in the modern educational process: Proceedings of the 6th Scientific and Methodological Conference, Minsk, March 24, 2009. Minsk: Belarusian State University; 2009. Pp. 55–58. (In Russian).

10. Tunda E.A., Tunda V.A. Human consciousness and digitalisation. System analysis in design and management. 2024;1:120–130. (In Russian). http://doi.org/10.18720/SPBPU/2/id24-27