By Teanna Barrett

February 25, 2026

Ethical AI requires the application of universal human values and international standards. However, it also needs to take into account Africa’s historical peculiarities. It needs to use a development paradigm lens to treat Africa’s social, economic and cultural rights on a par with more generalised civil and political rights. — Recommendations on the Inclusion Sub-Saharan Africa in Global AI Ethics, Arthur Gwagwa, 2019

The rise of “Big Data” and artificial intelligence (AI) is heralded as a leap forward in productivity, wealth, and living standards, yet these gains remain concentrated among a select few nations and within a privileged set of their populations. Terms like data colonialism and technofeudalism aim to conceptualize the newest iteration of power asymmetries and exploitation that power the Fourth Industrial Revolution (4IR). Yet these terms fall short of capturing the reality of the Global Majority. Like chattel slavery, colonialism, and imperialism, every major global leap in technology-mediated economic development has involved the exploitation of Africa and its diaspora. As with past industrial revolutions, African scholars and communities decry the exploitation of their labor, resources, and land by global superpowers. The inhumane working conditions of AI content annotators, the conflict-ridden extraction of raw minerals for computing hardware, limited access to electricity and computing infrastructure, and imbalanced collaborations with tech powers reveal the persistence of neocolonial dynamics in the global AI ecosystem.

To break the cycle of unequal development, African AI practitioners and ethicists have called for self-determined AI development — an approach that enables African communities to join the 4IR on their own terms. Drawing on decolonial and indigenous AI work and scholarship, data self-determination is a component of data sovereignty, in which indigenous communities exercise ownership over their data and have control over how that data is used. African data science communities also recognize that building the technology they imagine requires tackling political realities that hamper local AI development. The process of developing local AI ecosystems — self-owned and self-controlled — may offer a path toward sustained self-determined progress in the 4IR. Examining colonial projects of modernization, post-independence visions of African socialism, and present-day debates on governance reveal the enduring geopolitical logics that link 20th century development studies to AI development discourses today. Self-determined AI and data technology development in Africa requires decentering external assessments of success, breaking exploitative relationships with external tech powers, and committing to socially-conscious collectivist decision-making.

In “Development Theory and Changing Trends in Sub-Saharan African Economies 1960-89,” his chapter in the book African Perspectives on Development, Benedict S. Mongula notes that the paternalistic beginnings of African development discourses were characterized by externalized metrics of modernization. Colonial powers and global organizations such as the International Monetary Fund and the World Bank assumed the role of defining and setting the terms of development. To these institutions, development strictly meant economic viability in the global landscape. Performance metrics and export data were used to prove African economies as “backwards” due to their communalist economic systems. The assessing institutions stressed the need for investments and oversight from global powers to correct African economic trajectories. Especially as African nations sought independence, these investments came with specific conditions set by the lenders, which often involved coercing the production of cash crops to meet Global Minority demand. However, when poor environmental conditions, economic downturns, and civil unrest intensified in the late 20th century, the same cash crops African nations were encouraged to produce fell in value. The external development investors were relatively unaffected, but this market shift halted the economic growth of African nations. 

AI readiness reports administered by Global Minority institutions often reflect the same modernization logics of early development initiatives. These reports often place African countries near the bottom and inform the conditions in which tech powers and Global Minority institutions invest in African AI initiatives. But external AI development perspectives should not supersede and set the priorities for local AI development. When data are the cash crops of the 4IR, orienting data collection to serve global diversity representation needs places African AI development in a vulnerable position. As such, efforts to develop African data repositories should be led, owned, and meet the data discovery interests of local data providers.

Given the harms of early modernization development, African socialists and political leaders (including Julius Nyerere of Tanzania, Amilcar Cabral of Guinea-Bissau, Patrice Lumumba of Congo, Kwame Nkrumah of Ghana, and many more) called for independence from European colonialists, pushing for African societies to determine their own terms of development. In his 1972 book How Europe Underdeveloped Africa, Walter Rodney defines development in simple universal terms: the harnessing of environmental knowledge to build tools and improve how societies realize their destiny. He also argues the beginning of capitalism, with its focus on the production of raw materials to feed the unmitigated consumption of the Global Minority, violently siphoned opportunities for African nations to develop. Therefore, he asserts — and many African socialist thinkers agree — the only way for Africa to develop is by breaking dependent and extractive relationships with the Global Minority.  

The power asymmetries highlighted in 20th century decolonial critique persist today. African data ethics scholarship clearly shows that the AI ecosystem perpetuates colonial logics and constructs neocolonial power asymmetries epistemically, socially, and economically. AI development in Africa is held back by the exploitative and extractive relationship with tech powers as the metropoles and Global Majority technologist communities as the periphery. Decolonial and self-determined AI in Africa requires conscious activity, which Rodney defines as “grappling with the heritage of objective material conditions and social relations to enact a better status quo.” Towards conscious activity, grassroots data science collectives are showing real potential to challenge power and design local knowledge and alternatives. For example, Masakhane is a grassroots research collective for African natural language processing projects. In recognition of the legacy of extractive research, they explicitly discourage “parachute research” in which researchers from the metropole enter African research communities and use findings to benefit their interests with little benefit to the local community. Resetting the terms for AI research collaborations in Africa is a step towards dismantling colonial hierarchies in the new technology age.

While decolonial conceptions of African development laid the intellectual groundwork for independence, scholars emphasize that it was the collective struggle —  especially of peasant and working class Africans — that brought widespread decolonization to fruition. Yet in the 21st century, approaches to development have lost the transformative collectivist spirit. Philosopher Paulin Hountondji argues that post-independence African governments’ consolidation of power undermined and disempowered civil society. As a result, national agendas were not informed by collective knowledge and dismissed the practical needs of the majority. Today, African scholars are calling for a recommitment to communalist decision-making, so development can benefit the majority rather than just an elite few. 

The larger recommitment to African communalism is also found in recent regional AI organizations like Deep Learning Indaba. Clearly, class distinctions inform who is equipped with technical skills, in contact with politicians to develop AI policies, or able to secure the resources to fund large-scale projects. Recognizing classism in AI work is a prerequisite, but building a technology culture that rejects top-down agendas is crucial to sustainable progress. A self-determined and collectivist AI community builds mechanisms of accountability so practitioner decisions are informed, calibrated, and answerable to the perspectives of data contributors, local users, and impacted communities. If community members do not want to use data technology at all, design considerations should enable both users and non-users to participate equally in society. Many of these organizational structures, such as resource stewardship and consensus-based decision-making, are already practiced within various indigenous African governance structures. 

The contemporary obstacles to self-determined progress in the 4IR aren’t insurmountable. Through a clear understanding of material conditions and resolute collective action, African societies have been able to disrupt the status quo in their best interests. As Rodney argues in the final section of How Europe Underdeveloped Africa, it’s the continual struggles — and the work of connecting these struggles — that build momentum and lead to larger-scale change. In the realm of democratizing AI, innumerable communities are doing their part. There is also potential to build solidarity across the African diaspora through recognizing our shared approaches to democratizing AI. In the United States, for example, opposition to the development of data centers in Black communities and creating community-owned African American English (AAE) datasets mirrors similar environmental sustainability and language-preservation efforts on the continent. Democratizing AI in Africa isn’t just about building sovereign AI, but working toward the political, economic, and social conditions to nurture self-determined AI development.

References

Chachage, C.S.L. (1994). Discourse on Development among African Philosophers. In U. Himmelstrand, K. Kinyanjui & E. Mburugu (Eds.), African Perspectives on Development (pp. 84-95). St. Martin’s Press.

Mongula, B.S. (1994). Development Theory and Changing Trends in Sub-Saharan African Economies 1960-89. In U. Himmelstrand, K. Kinyanjui & E. Mburugu (Eds.), African Perspectives on Development (pp. 84-95). St. Martin’s Press.

Rodney, Walter (1981). How Europe Underdeveloped Africa. Howard University Press.

Teanna Barrett is a second year PhD student at the Paul G. Allen School of Computer Science & Engineering at the University of Washington. She earned her B.S. in Computer Science with a minor in Philosophy from Howard University. Her current research aims to understand how data scientists form social analyses to motivate their work and translate these motivations into technical practices (and vice versa). Towards this inquiry, Barrett engages with Black thought, data ethics, and human-centered design techniques to imagine tools to support the development of data science praxis. She most recently presented her paper, “African Data Ethics: A Discursive Framework for Black Decolonial Data Science” at the 2025 ACM Computing Conference on Fairness, Accountability, and Transparency (FAccT).