Artificial intelligence use and scientific innovation

Abstract The rapid advancement of artificial intelligence (AI) technologies is profoundly reshaping scientific research and accelerating its progress. While prior studies have explored AI applications across various disciplines, an understanding of whether AI use contributes to scientific innovation remains limited. In this study, we propose the AI use score, a novel metric that quantifies the extent of AI use in scientific research. It includes two key dimensions: a term‐based score, derived from the presence of AI‐related terms in titles and abstracts, and a knowledge‐based score, based on citations to AI‐related literature in reference lists. Our findings reveal that AI use is positively and significantly associated with both scientific disruption and novelty, with this relationship being particularly pronounced in STEM disciplines. The association between term‐based AI use and scientific innovation has steadily intensified over time. Notably, combining both dimensions of AI use shows the strongest correlation with innovative outcomes. Specifically, term‐based AI use is more strongly linked to disruptive innovation, while knowledge‐based AI use is more closely associated with scientific novelty. Furthermore, we uncover the mechanism of how AI use influences scientific innovation by exploring the text similarity of publications.