Modern society is experiencing a digital transformation of various spheres associated with the development of artificial intelligence and big data. The introduction of large language models (hereinafter referred to as LLM) into scientific research opens new opportunities, but also raises a number of questions, which makes it relevant to study the peculiarities of their application for qualitative data analysis in sociology. The purpose of this article is to explore how LLM can influence the methodology and practice of sociological research, and to identify the advantages and disadvantages of their application. The authors rely on the use of the Calude-3 LLM to qualitatively analyse empirical data from a sociological study of students’ perception of entrepreneurship. The possibilities of LLM in the analysis of qualitative data are revealed: analysis of sentiment, construction of logical conclusions, classification, clustering, and formation of typologies. The advantages of using LLM are shown: increased data processing speed, saving time and resources. The application of LLM becomes a tool to optimise the research process in sociology, allowing to deepen the analysis of qualitative data, but it also has a number of limitations: social and political bias, difficulties with hallucinations. It is necessary to increase the transparency of models, improve their interpretability and explainability and reduce their social and political bias as well as ethical and legal regulation of the use of LLM models.