The review, led by Ayodele Samuel Jegede with Olayinka Akanle and Olusegun Israel Olaniyan, sets out to answer a deceptively simple question: why does the sociology of AI matter in Africa, and what historical and social contexts shape the technology’s incursion into the continent? The authors’ answer is that AI does not arrive in a vacuum. Like every technological innovation before it, it lands within pre-existing social systems, colonial legacies and structural conditions that determine who benefits, who is excluded and whose knowledge counts.

Most existing scholarship on AI, the authors argue, concentrates on the technology’s potential implications for everyday human society in general terms, leaving a gap in understanding how AI’s development and utilisation reflect and reshape existing social structures within African societies specifically. Earlier work has touched on the sociological implications of AI for inequalities, power and data justice, but the authors contend that such studies lack in-depth theoretical explanation and pay limited attention to the asymmetrical relationships inherent in everyday AI use in Africa. Their remedy is to approach the subject from a decolonial perspective, asking what happens when AI systems designed with systematised Western data structures encounter African informal social realities, and which forms of African knowledge are excluded because they are not technologically readable.

Methodologically, the article is anchored on a purposive collection of secondary documents in three categories: peer-reviewed journal articles on AI, the sociology of AI and African development; institutional and policy documents, principally the Kenya AI Strategy 2025–2030; and an industry report on artificial intelligence in Sub-Saharan Africa produced by Google, Sand Technologies and African Leadership University. Documents were sourced from platforms including Google Scholar, ResearchGate, IEEE Xplore and JSTOR, using keyword combinations such as “artificial intelligence,” “AI and digital divide,” “AI in Africa” and “sociology of AI.” The authors applied a Critical Reliability Assessment to filter the literature, restricting inclusion to studies published between 2012 and 2025, a period they describe as one of rapid proliferation of AI research in Africa, and excluding work focused purely on technical or computational aspects with no relation to African development.

The theoretical backbone of the review is decolonial theory, which holds that the world and what counts as knowledge have been constructed on a European ontological perception of modernity, erasing the worldviews, institutions and knowledge of less privileged societies. Drawing on the Peruvian sociologist Anibal Quijano’s concept of the coloniality of power, the authors describe how control of economy, authority, gender and subjectivity continues to structure postcolonial societies. Applied to AI, this framework suggests that the technology reproduces and accentuates established social inequalities, including the digital divide and racial and economic disparities, because the substantial inputs AI requires, such as data and human skills, are unevenly distributed and shape what is even available in Africa.

The authors argue that asymmetrical relationships between Africa and the developed countries of the Global North continue to demonstrate and perpetuate historical domination, technological overpowering and epistemic exclusion. Most existing and strategic technological infrastructures, skills, tools, demands, markets and knowledge production relevant to AI are determined and moderated by external agencies, interests, actors and institutions, a pattern they trace back to the colonial era. Even in the few African countries with growing AI research and capabilities, including Kenya, Nigeria and South Africa, the authors state that research, funding, demands and capabilities are driven from outside the continent, demonstrating continued dependence on global dictates. On this reading, AI is expanding and reconfiguring unbalanced technological power relations against Africa, and will continue to do so unless AI is decolonised by decentring its technological inputs and incorporating more diverse, organic knowledge systems.

The review also traces the historical arc of the field itself, from Alan Turing’s 1950 question of whether machines can think and the 1956 Dartmouth Conference where John McCarthy coined the term artificial intelligence, through the symbolic AI and expert systems of the 1980s, to the machine learning and deep learning era from the 1990s onward. Against this long Global Northern timeline, the authors note that the roots of AI in Sub-Saharan Africa, as documented in the 2025 industry report, date back only to the early 2000s, when academic institutions began establishing computational science research programmes. The African Institute for Mathematical Sciences, founded in 2003, began offering programmes in machine learning and data science, and the launch of the IBM Research Lab in Nairobi in 2013 marked a significant investment in indigenous AI capabilities. By 2023, the report counted more than 2,400 AI-focused companies and tech hubs in countries including South Africa, Nigeria and Kenya, and governments such as South Africa’s, with its Presidential Commission on the Fourth Industrial Revolution, and Kenya’s, with its AI task force, have begun building policy frameworks around adoption.

The review documents concrete applications alongside persistent gaps. In Nigeria, the HelpMum ADVISER framework, an AI-driven vaccination intervention optimiser, has been deployed to more than 13,000 families in Oyo State to optimise vaccination programmes, illustrating how AI can support public health interventions in resource-constrained environments. Yet the authors also cite a quantitative study of healthcare students at Obafemi Awolowo University in which, while 60 percent of respondents believed they possessed high levels of AI skills, objective assessment found that approximately 92 percent demonstrated limited or low skill levels, with only 0.4 percent achieving high knowledge scores. Only 12 percent of respondents correctly defined machine learning, 7.5 percent understood supervised and unsupervised learning, and 10.2 percent grasped Internet of Things concepts. The authors flag this gap in understanding training data bias, model validation and algorithmic transparency as particularly problematic for safe and effective AI implementation in healthcare, compounded by inadequate infrastructure and weak ethical and regulatory services.

Kenya serves as the article’s case study in how an African country is navigating these tensions. Its AI strategy aspires to make the country a regional leader and global force by building an inclusive AI ecosystem that leverages indigenous information and talent, prioritises data sovereignty, and aligns with global human rights standards. The authors, however, offer a pointed critique of the strategy’s own framing: Kenya is positioned as an aspirant and adopter rather than an originator of AI paradigms, and responsible AI is defined largely through alignment with global standards rather than Kenyan or African normative frameworks. In their reading, the strategy’s discourse reproduces the very asymmetrical power relationship its decolonial framing seeks to critique, asserting sovereignty at the level of talent and data while anchoring legitimacy externally. Kenya also faces a shortage of AI experts capable of training contextually relevant models, lacks a holistic regulatory framework specifically targeted at AI, and struggles with insufficient computing ability and poor data quality, with digital divides between urban and rural areas and marginalised communities hindering adoption.

The authors conclude that AI in Africa carries genuine potential to address long-standing development challenges, driving innovations in financial inclusion, climate resilience, agriculture and food security, rapid diagnosis, education, language inclusion and job creation, while simultaneously exacerbating technological asymmetry, expanding existing inequalities and perpetuating historical institutional dynamics. Their recommendations centre on co-design and participatory development with communities, indigenous knowledge holders and marginalised populations; context-aware and flexible regulatory frameworks; capacity building through education and infrastructure; and ethical pluralism, including indigenous philosophies such as Ubuntu as normative foundations for AI governance. They call for grounded socio-technical and ethnographic research on how AI systems actually function in African contexts, and for Africa to be treated as a necessary AI partner rather than a mere consumer. Because the article is a review based on secondary documents and three illustrative national cases, the authors themselves caution against generalising its conclusions to the entire continent, but its central claim stands: without a sociologically informed approach attentive to history, power, culture and inequality, AI’s transformation of Africa will remain unevenly shared.

Subject of Research: A decolonial sociological review of how artificial intelligence is developed and used in Africa, including a case study of Kenya's AI strategy.

Article Title: The sociology of artificial intelligence development and utilisation in Africa

Article References: Jegede, A. S., Akanle, O., & Olaniyan, O. I. (2026). The sociology of artificial intelligence development and utilisation in Africa. Frontiers in Sociology, 11, 1917558. https://doi.org/10.3389/fsoc.2026.1917558

Image Credits: AI Generated

DOI: 10.3389/fsoc.2026.1917558

Keywords: artificial intelligence, Africa, sociology of AI, decolonial theory, digital divide, Kenya AI strategy, coloniality of power, healthcare AI, data sovereignty, technology and inequality, sociology, artificial