Scientific knowledge has long relied on the hypothetico-deductive method, rooted in observation, experimentation, and causal inference (Velthoven & Marcus, 2024). With formal logic (Boole, Frege) and early computing (Turing), AI emerged as an extension of rational processes (Feng, 2019).<\/p>\n
Today, machine learning shifts this paradigm, favoring data-driven correlations over deductive rules (Alvarado, 2023). This raises new epistemological challenges about the nature, validity, and explainability of AI-generated knowledge (Bao & Zeng, 2024).<\/p>\n
To epistemologically assess the limits and scope of the knowledge generated by AI models, considering the difference between correlation and causality, as well as the problem of explainability. Using as a starting point a database of articles from the last three years in a bibliographic repository, highlighting universities and journals interested in studying the method in the era of AI.<\/p>\n
From an initial set of 700 articles, 55 met the strict criteria of addressing both epistemology or philosophy and artificial intelligence. The UK led in publication output, and all articles were published in English. Of the authors, 37 were men and 18 women, highlighting ongoing gender disparities and the dominance of English in academic discourse.<\/p>\n
The peak year for publications was 2024. These trends suggest a concentration of epistemic authority in the Global North, echoing Dussel\u2019s (1996) call to question the persistence of geopolitical centrality in knowledge production.<\/p>\n