From data scarcity to data abundance: big data, artificial intelligence and the transformation of knowledge production in the social sciences
DOI:
https://doi.org/10.26577/JPsS20994926Abstract
The article examines the development of data-driven research in contemporary social sciences amid the growing use of big data, machine learning, and artificial intelligence. The aim of the study is to analyze publication trends, the thematic structure of this research field, and the prevalence of critical discourse concerning the methodological and epistemological limitations of digital technologies. The empirical dataset comprises 3,779 English-language scientific articles published between 1995 and 2025 and indexed in the Scopus database. The study employs bibliometric analysis, natural language processing (NLP), latent Dirichlet allocation (LDA) topic modeling, bigram analysis, and dictionary-based identification of critical themes. Particular attention is given to data quality, measurement validity, algorithmic bias, and risks associated with generative artificial intelligence. The findings demonstrate substantial growth in publication activity since the 2010s and a shift in thematic emphasis from big data toward machine learning and artificial intelligence. Topic modeling reveals considerable thematic heterogeneity despite the application of similar analytical approaches. Indicators of critical discourse are identified across different thematic areas, although their prevalence remains limited: relevant publications accounted for 9.2% of the annual dataset in 2024 and 9.0% in 2025. The findings indicate that quantitative research expansion is accompanied by thematic fragmentation and a relatively small proportion of publications containing identifiable indicators of critical reflection. The scientific significance of the study lies in systematizing research trends and thematic structures, while its practical significance concerns improving social research methodology and teaching data analysis methods. However, term-frequency analysis cannot directly assess the depth of critical reflection or the degree of theoretical integration within the social sciences. This limitation should be considered when interpreting the findings and drawing broader conclusions.
Keywords: big data, artificial intelligence, datafication, machine learning, critical reflection, digitalization.









