Decolonial AI: Algorithmic Colonialism and the Struggle for African (Technological) Futures

This article examines artificial intelligence (AI) as a contemporary formation through which colonial epistemologies, infrastructures, and imaginaries can be reproduced and reconfigured. Drawing on coloniality, data colonialism, algorithmic racism, and decolonial computing, we argue that dominant forms of AI development can extend historical logics of extraction, classification, and domination into digital and predictive systems. Centering Africa and the Global South, the article examines how these relations operate across four interconnected domains: the coloniality of power, being, knowledge, and ecology. Through a decolonial critical-theoretical approach supported by illustrative cases, it traces how AI can reproduce asymmetries in infrastructure ownership, labor, computational recognition, knowledge production, and environmental costs. The article argues that AI bias extends beyond technical deficiencies that can be addressed through inclusion or representational diversity. It is also shaped by the historical and material conditions of colonial modernity, through which particular modes of knowledge are universalized while others are marginalized or rendered computationally illegible. However, AI does not produce uniform effects across contexts and can be appropriated to support alternative forms of knowledge production and technological agency. The central concern, therefore, is how the epistemic, political-economic, and infrastructural conditions of AI might be transformed. The article examines decolonial AI as a contested project encompassing refusal, strategic engagement, and sociotechnical reconfiguration. Drawing on African AI initiatives and debates on technological sovereignty, it develops a pluriversal orientation that connects epistemic plurality with institutional transformation. It proposes that decolonial AI requires interventions across governance, data, labor, and infrastructure, alongside greater capacity for African institutions and communities to shape technological development. Rather than replacing one universal technological order with another, the paper argues for technological futures in which multiple ways of knowing, living, and organizing social life can participate in determining the conditions of AI.