paper

Decolonial AI

A critical framework examining how artificial intelligence systems reproduce colonial power structures and proposing decolonial approaches to AI development and deployment.

Type
paper
Year
2020
By
Shakir Mohamed et al.

Decolonial AI is a critical framework that examines how artificial intelligence systems, from training data to deployment contexts, reproduce and amplify colonial power structures, epistemic hierarchies, and extractive relationships. The framework draws on decolonial theory and postcolonial scholarship to interrogate whose knowledge is encoded in AI systems, whose labor is exploited in their creation, and whose interests they serve. Proponents argue that conventional AI development often treats Western technical rationality as universal while marginalizing non-Western knowledge systems, indigenous perspectives, and the Global South's agency. The decolonial AI approach proposes alternative development practices including participatory design with affected communities, centering indigenous data sovereignty, redistributing AI benefits, and building AI systems that respect cultural pluralism and challenge rather than reinforce hierarchies.

Last updated 31 August 2026