GENERATIVE ARTIFICIAL INTELLIGENCE IN SCIENTIFIC RESEARCH: A QUANTITATIVE SOCIAL SCIENCE PERSPECTIVE

Abstract

Generative artificial intelligence is increasingly transforming scientific research by supporting idea generation, methodological design, literature search, data interpretation, and academic writing. This study examines the use, perceived benefits, and limitations of generative AI tools in scientific research from a quantitative social science perspective. Based on Innovation Center University of Niš survey data collected from 100 respondents, the findings indicate that a substantial majority of participants have already used generative AI tools in their research activities. Specifically, 82% of respondents reported previous use of generative AI, while 18% had not used such tools. Among users, the most frequently reported reasons for using generative AI included searching the literature with similar methodological frameworks, generating initial research ideas, identifying possible shortcomings and corrections, and obtaining consultative support in methodological decision-making. Respondents who had not used generative AI most often cited insufficient reliability, ethical concerns, limited competence in using AI tools, and the absence of perceived need at the current stage of research. The results further suggest that researchers generally perceive generative AI as a tool that can contribute to the efficiency of scientific work. On a five-point scale, the majority of respondents expressed moderate to high agreement with the statement that generative AI can improve research efficiency. In terms of the methodological framework, the greatest perceived benefits were identified in selecting specific methods and instruments for research implementation, generating research writing ideas, defining research objectives, and formulating hypotheses. These findings indicate that generative AI is not primarily viewed as a replacement for researchers, but as an auxiliary tool that can enhance productivity, support methodological reasoning, and improve the organization of research tasks. The study contributes to the emerging discussion on the responsible integration of generative AI in scientific research, particularly within the social sciences. It highlights both the practical potential of AI-assisted research and the need for methodological literacy, ethical awareness, and critical evaluation of AI-generated outputs. The paper concludes that generative AI can significantly support scientific research when used transparently, critically, and in accordance with established academic standards.