For generative Artificial Intelligence (AI) to truly succeed in African higher education, the continent requires approaches tailored to its unique context rather than merely adopting solutions developed elsewhere. Although students are already integrating GenAI into their studies, universities and governments are still in the process of formulating regulatory policies. Addressing this gap requires navigating significant challenges, such as limited digital access, linguistic diversity, low tertiary participation, and the underrepresentation of African cultures within global AI systems. These insights are drawn from a paper titled: GenAI Adoption in Higher Education Across the Global South: A Call to Extend the Evidence Base to Africa which utilised mixed-methods approach including surveys of 1,523 students across 17 countries and 32 interviews with academics to examine the specific drivers and barriers affecting the effective adoption of GenAI in the region.
Student trust and continued use
The study indicates that students' intentions to continue using generative AI are primarily driven by trust and perceived ease of use, with those intentions strongly predicting actual usage. While academics report positive experiences, particularly regarding research efficiency and time savings, these benefits are tempered by significant challenges. Specifically, nearly half of those interviewed cited infrastructure limitations and the digital divide especially in underserved areas alongside persistent ethical concerns involving data privacy, plagiarism, and the spread of misinformation.
Infrastructure and geographic disparities
The challenges hindering GenAI adoption in African higher education largely mirror global issues, such as unreliable connectivity, high costs, capacity gaps, inconsistent policies, and a reliance on external technologies. However, these familiar obstacles are compounded by specific regional dynamics, including geographic disparities, linguistic diversity, unequal access to higher education, and the marginalisation of African knowledge and cultures within global AI systems.
Higher education systems often span vast, diverse geographies, with many institutions located far from major urban centres. Consequently, for many students, academic access is determined not just by internet quality, but also by their physical proximity to campus, the availability of digital services, and the broader scope of institutional support.
Language and cultural representation
Language is a critical factor in the broader challenge of equitable generative AI adoption, particularly in Africa, where an estimated 1,500 to 2,000 languages exist, yet fewer than 1% are represented in global AI systems. This disparity creates a significant hurdle for students, who are often required to study in English, French or Portuguese while primarily thinking and communicating in their native languages. Consequently, the challenge extends far beyond the technical capacity for translation; the more pressing issue is whether these systems can meaningfully support students whose languages, knowledge systems and cultural contexts are fundamentally underrepresented in the data used to train AI models.
Equity and locally responsive design
Addressing the adoption of generative AI within African higher education requires consideration of two primary issues: first, whether currently enrolled students have equitable access to these tools, and second, whether these technologies will serve to bridge educational divides or instead exacerbate existing inequalities by favouring those already equipped with the necessary devices, connectivity, paid subscriptions, and digital literacy. This shift in perspective reframes the discourse from merely asking who is currently using generative AI to questioning who is truly positioned to benefit from it.
GenAI adoption cannot follow a one-size-fits-all model; it must prioritise solutions responsive to regional contexts, particularly in Africa, where global systems often suffer from a lack of representation regarding local languages, cultures, and knowledge. Because global models predominantly learn from existing digital data, African research and archives are frequently underrepresented, leading to potentially stereotyped AI-generated outputs. The critical challenge for African higher education is to ensure that students and educators can recognise their own realities within these technologies. This perspective does not reject GenAI but advocates for a nuanced understanding of its limitations, proposing that co-design is essential to prevent the imposition of external assumptions, ultimately allowing African institutions to produce research that is comparable in rigour but uniquely authentic in its framing.
_________________________
Article Editor: Moloko Mathipa-Mdakane
Comments