The Political Economy of AI Development
This paper examines how the material conditions of Artificial Intelligence production create new forms of technological dependency between the Global North and South. While current discussions of AI democratization often focus on data accessibility, this analysis reveals how the concentration of computational infrastructure and specialized technical labour in a handful of Northern countries reinforces core-periphery relations in the global technological order.
The research analyses three key aspects of AI production that perpetuate North-South power dynamics: (1) the massive computational infrastructure requirements that create insurmountable barriers to entry for Global South institutions, (2) the transformation of Global South cultural heritage and knowledge into training data without meaningful participation in its governance, and (3) the way technical expertise and labour flows predominantly benefit Northern technology corporations, reflecting Cardoso’s concept
of ‘associated-dependent development’.
This transdisciplinary analysis, drawing from the epistemological intersection of political economy, philosophy, and STS, presents a detailed examination of current Artificial Intelligence development practices. The paper demonstrates how seemingly neutral technical decisions about model architecture, training data, and optimization metrics encode specific economic and political priorities that systematically marginalize Global South perspectives. The research draws particular attention to how recent proposals for ‘open data’ and ‘AI democratization’ often mask what Andre Gunder Frank identified as the ‘development of underdevelopment’ in the technological sphere.
The paper concludes by proposing alternative frameworks for AI development that centre Global South epistemologies and interests. These include new models of technological cooperation, alternative approaches to data governance, and mechanisms for ensuring more equitable distribution of AI’s benefits. By highlighting these structural inequalities in AI production, this research contributes to broader discussions about technological autonomy and creating more equitable forms of global technological development.
Read More ➜