The research, led by J. N. A. Aryee of Kwame Nkrumah University of Science and Technology in Ghana together with colleagues across five countries, combined two complementary strands of evidence. The qualitative strand consisted of multi-stakeholder focus group discussions in Ghana, Rwanda, and Namibia, each bringing together educators, learners, industry representatives, and policymakers in a single session to enable direct cross-sector dialogue. In total, 77 of 90 invited participants took part, with sessions of 27, 26, and 24 people respectively. The quantitative strand was a cross-country survey of 156 respondents spanning Ghana, Rwanda, Namibia, Kenya, and Zambia, harmonised onto a common ordinal scale and analysed with descriptive statistics and thematic coding of open-ended answers. The study was grounded in three theoretical frameworks: the Technology-Organisation-Environment framework, the Culture-Centred Approach, and Socio-Technical Systems theory, the last of which supplied the study’s central explanatory principle that technological integration fails when only the technical or only the social subsystem is addressed.

The focus group findings revealed three dramatically different national trajectories. Rwanda operates under a strong top-down policy vision, with an enacted National AI Policy and home-grown innovations such as the E-Ubuzima health application and the Sophia traffic robot, yet its universities remain hampered by institutional rigidity. Curriculum review cycles take roughly five years, a pace one University of Rwanda participant described as ill-suited to a rapidly evolving AI landscape, and many instructors themselves lack training in contemporary AI methods. Namibia, by contrast, exhibits pragmatic institutional adaptability, with accreditation cycles every four to five years and a 20 percent flexibility margin that allows emerging technologies to be woven into programmes without full curricular overhaul. But its small domestic market limits employment opportunities for specialised AI talent, and universities continue teaching tools such as SPSS while local industry moves toward Python, R, and machine learning libraries. Ghana demonstrates emergent, bottom-up innovation, exemplified by KNUST’s Responsible AI Lab, which trains high school students and runs staff summer schools, yet this proactivity operates within a national policy vacuum that leaves institutions without coherent direction.

Despite these divergent pathways, the study found that all three countries converge on the same underlying failure mode. In Rwanda, technical adoption across government and the private sector has outpaced the social subsystem of formal education, leaving universities without governance frameworks or faculty training adequate to the pace of change. In Namibia, institutional agility, a social strength, is undermined by technical stagnation in curricula. In Ghana, students and faculty adopting advanced AI tools are dangerously ahead of policy, pedagogical guidance, and ethical frameworks. The survey data corroborated this pattern at scale: awareness of AI’s strategic importance is high across all five countries, with contract staff averaging belief scores above 3.0 on the 0-to-4 scale, but engagement in AI-related coursework averaged just 1.43 among students and participation in AI activities only 0.98, both within the occasional range. Recognising AI as transformative, the authors conclude, and being institutionally equipped to develop its capacities are deeply disconnected conditions.

The survey also exposed a striking paradox at the heart of curriculum design. Among students who rated their AI courses as very relevant to industry demands, only 28 percent agreed they felt prepared for the job market, while 41 percent remained neutral and 22 percent disagreed. Even well-aligned course content, in other words, does not translate into confidence without experiential learning, applied projects, and structured industry engagement. Country-level curriculum priorities varied considerably: machine learning dominated in Ghana, Kenya, Namibia, and Rwanda, while Zambia showed a striking concentration on computer vision at 66.7 percent, possibly reflecting sectoral demands in agriculture, surveillance, or medical imaging. Rwanda displayed the most diversified profile, with artificial neural networks and computer vision complementing machine learning.

Equity emerged as the most critically under-addressed dimension of AI integration. Only 18 percent of survey respondents felt fully included in AI-related education, while 23 percent reported feeling excluded, and targeted initiatives for underrepresented groups were reported by just 22 percent. Financial constraints, cited by roughly half of respondents, formed a multidimensional barrier spanning tuition, certification costs, internet data charges, transport, and devices. Infrastructure limitations affected more than a third, and awareness gaps left nearly half unaware of available programmes. Women made up only 35.8 percent of survey respondents, and the focus groups documented compounding barriers for women, persons with disabilities, rural students, and part-time workers, from the absence of Kinyarwanda-language AI content in Rwanda to unaddressed accessibility barriers in Ghanaian AI tools and learning platforms. The Culture-Centred Approach, however, also surfaced expressions of agency: Ghana’s emphasis on data sovereignty and homegrown platforms, Rwanda’s demand for localised content, and Namibia’s call for digital education beginning in primary school all constitute implicit frameworks for locally owned solutions.

University-industry collaboration was identified as the single most important enabler of better AI education, prioritised by 76 percent of survey respondents, followed by better computing facilities at 69 percent, scholarships at 68 percent, and hands-on projects at 67 percent. Yet the current state of these linkages falls far short. In Rwanda, internships are predominantly unpaid, informal, and inaccessible to disadvantaged students. In Namibia, pilot partnerships with BMW and Bloomberg Finance are promising but limited in scale. In Ghana, stakeholders reported that universities and industries each lack clarity on what the other wants, with limited mechanisms for industry to communicate skill needs back to academia. The study calls for a structural architecture of collaboration encompassing co-designed curricula, jointly supervised applied research, compensated internships, and co-teaching arrangements with industry practitioners.

Governance emerged as the missing enabler underlying all other findings. In Rwanda, a National AI Policy exists but institutional adoption is inconsistent, creating what stakeholders described as a vacuum of regulation. Namibia lacks a cohesive national AI strategy entirely. Ghana operated within a significant policy vacuum despite institutional proactivity. Most tellingly, 69 percent of survey respondents were unsure whether their country even has a national AI policy, and only 4 percent described themselves as very familiar or expert in AI-related regulations. Most respondents reported that their institutions had not adopted AI ethics guidelines or responsible AI frameworks. The authors argue that the governance challenge is not merely whether policies exist but whether they are communicated, translated into institutional frameworks, and supported by implementation mechanisms, with the African Union’s 2024 Continental AI Strategy providing an important but under-penetrating regional framework.

The study’s recommendations are calibrated to four sets of actors. Higher education institutions should establish institution-wide AI policies, formal university-industry curriculum committees, and interdisciplinary AI working groups, while embedding AI ethics and locally relevant datasets into core learning outcomes. National governments should move beyond policy statements to resourced implementation plans, invest in national data infrastructure as a public good, create incentives such as matched funding and tax relief for private-sector participation, and set dedicated equity targets with measurable indicators for rural connectivity and participation of underrepresented groups. Industry partners are urged to become active co-investors rather than passive consumers of talent, formalising advisory roles and expanding compensated internships with measurable inclusion targets. Regional bodies should harmonise AI education standards and build shared infrastructure, from regional computing facilities to secure data repositories, through platforms including the African Union strategy, ECOWAS, and SADC.

The authors acknowledge limitations, including the modest purposive sample of 156 survey respondents, uneven stakeholder representation across countries, and the snapshot nature of data collected in a rapidly evolving landscape, meaning findings should be read as exploratory rather than nationally representative. Yet the core conclusion stands with unusual clarity across both methods: sustainable AI integration in African higher education is not primarily a technical challenge but a socio-technical one. Technical advancement that outpaces social, pedagogical, and governance co-evolution does not produce progress; it produces, in the authors’ words, a new and more technologically sophisticated form of the same structural inequalities. For a continent whose youthful population represents a potential demographic dividend for AI-driven development, the stakes of getting this co-evolution right could hardly be higher.

Subject of Research: AI integration, equity, and governance in sub-Saharan African higher education

Article Title: Rethinking AI integration in African higher education: a mixed-methods investigation of capacity, equity, and governance across sub-Saharan Africa

Article References: Rethinking AI integration in African higher education: a mixed-methods investigation of capacity, equity, and governance across sub-Saharan Africa. (n.d.). https://doi.org/10.3389/feduc.2026.1968304

Image Credits: AI Generated

DOI: 10.3389/feduc.2026.1968304

Keywords: artificial intelligence, higher education, sub-Saharan Africa, digital equity, AI governance, RetAIn project, curriculum reform, university-industry collaboration, socio-technical systems, Ghana, Rwanda, Namibia