Understanding AI and power: situated perspectives from Global North and South practitioners

Global debates on artificial intelligence (AI) ethics and governance remain dominated by high-income, AI-intensive nations, marginalizing perspectives from low- and middle-income countries and minoritized practitioners. This qualitative study adopts a decolonial and sociotechnical lens to examine how AI practitioners across Africa, Asia, South America, the Caribbean, and minoritized groups working in high-income contexts conceptualize AI's value, harms, and governance. Drawing on reflexive thematic analysis of 22 in-depth interviews, the study explores how geographic, cultural, and professional contexts shape practitioners' understandings of ethics, harm, and power within the global AI ecosystem. Findings reveal a dual orientation. While some participants view AI as a neutral tool shaped by human intent, others frame it as a sociotechnical system that reproduces structural inequities through data colonialism, exclusion, and epistemic dependency. Despite these asymmetries, participants articulated cautious yet agentic imaginaries of AI's potential to address local and regional problems in healthcare, education, and public governance. The study advances decolonial AI ethics by empirically grounding how ethical reasoning and governance are negotiated under constraint and by highlighting pathways toward more equitable, context-sensitive global AI governance. Supplementary Information: The online version contains supplementary material available at 10.1007/s00146-025-02731-x.