Rong Ni, Yao Qu, Jue Wang
Abstract Artificial intelligence (AI) has become a strategic infrastructure for contemporary science, raising pressing questions about whether it will reinforce existing hierarchies or foster more inclusive knowledge production. This study analyzed 134,719 AI-assisted articles using a large-scale large language model-based approach to examine spatial and temporal disparities between the Global North and the Global South (excluding the US and China). Results reveal several global patterns: convergence is observed in research output, while disparities persist in journal prestige. Substantial disciplinary variation also emerges, with digitally accessible fields, such as physics and technology, providing comparatively more equitable entry points for Global South researchers. Notably, collaboration networks exhibit clear North–South asymmetries. Joint publications involving major AI powers such as the US or China are consistently associated with higher journal prestige for the North and South. These findings suggest that AI-assisted research broadens participation and entrenches structural hierarchies, highlighting the need for governance and policy interventions that promote equitable access and inclusive collaboration.