Research Hub
The intellectual center of AIRSA — publications, white papers, and policy briefs.
Strengthening Artificial Intelligence Education in Ghana
Ghana has set the right national direction on AI education, but its teachers, schools and safeguards are not yet ready to deliver it. Closing that delivery gap within three years is the most important step in turning the National AI Strategy into real skills for learners. The National AI Strategy 2025–2035, launched on 24 April 2026, treats education and workforce readiness as a core pillar. A revised basic school curriculum with coding and foundational AI literacy has followed, and subject-specific AI apps are being rolled out in senior high schools. These are real advances, and few African systems have moved as quickly. However, an education-sector review of the strategy finds it stronger on national AI readiness than on school-level implementation. Teacher preparation, assessment guidance, learner data protection, local-language tools and rural infrastructure remain thin. Without action, AI education risks widening the gap between well-resourced urban schools and everyone else. This brief recommends eight actions for the Ministry of Education, the Ministry of Communication, Digital Technology and Innovations (MoCDTI) and their agencies: • Publish a dedicated AI-in-Education implementation plan, funded from the National AI Fund. • Make teachers the centre of reform, from colleges of education to in-service training. • Build curriculum progression and assessment rules from kindergarten to university. • Set binding safeguards for learner data and AI tools used in schools. • Close the infrastructure and equity gap, starting with offline-capable tools. • Invest in Ghanaian-language and locally grounded AI tools. • Speed up AI programme approval in universities and embed AI skills in TVET. • Govern the effort through a standing, multi-stakeholder AI-in-Education unit with public indicators.
James Mane · 2026 · Ghana
Digital Technology Adoption and Agricultural Productivity among Smallholder Farmers in Africa: Barriers, Benefits, and Policy Implications
Smallholder farmers produce much of Africa's food, yet their yields and incomes remain low, and digital technologies are widely promoted as a way to close the gap. This paper reviews the evidence on how digital technology adoption affects smallholder productivity in Africa, what prevents adoption, and what policy should do. It sets out a four-step framework (access, sustained use, behaviour change, economic outcome) and separates evidence on adoption from evidence on impact. The review finds that benefits are real but modest. A 2025 meta-analysis of phone-mediated agricultural technology scaling found pooled gains of about 2% in yields and 5% in profit, with adoption much higher when phones were combined with traditional extension; a Kenyan SMS trial reported 11.5% higher yields. Mobile phone rollout in Mozambique reduced maize price dispersion by 4.5% to 7%, but the share reaching farmers is unclear, and information did not raise farm-gate prices in Niger. Equipment-sharing platforms and AI advisory such as Farmer.Chat show strong usage but no measured impact yet, with a 600-village trial in Kenya running to 2027. The main barriers are the cost of data and handsets, a 29% mobile internet gender gap in sub-Saharan Africa, limited skills and trust, gaps in local-language content, and missing credit, inputs and competitive markets. The paper recommends blending digital and human extension, making access affordable, designing for women and the poorest, investing in local content, bundling information with inputs and finance, protecting farmers' data, and requiring impact evaluation.
James Mane · 2026 · Africa
Artificial Intelligence and Poverty Reduction in Africa: Assessing the Pathways to Inclusive Economic Growth
Artificial intelligence (AI) is widely expected to help African countries reduce poverty and grow more inclusively, but the evidence for this is uneven. This paper reviews six pathways from AI to poverty reduction in sub-Saharan Africa (agriculture, health, financial inclusion, education, public services, and jobs and enterprise) and assesses the African evidence for each. It sets out a framework with four channels (productivity of the poor, access to services, employment, and public resources) and tests each pathway against whether it reaches people below the poverty line, lowers costs, shows measured outcomes and shares value locally. The review finds that the strongest evidence lies in education (a randomised trial of an AI maths tutor in Ghana found a gain of 0.36 standard deviations at about US$5 per student) and in clinical decision support in Kenya (diagnostic errors down 16% across nearly 40,000 visits). Evidence is weaker for agriculture, AI-enabled credit and public services, and the quality of entry-level AI jobs is a concern. No pathway yet has proof of lasting income gains for poor households. The binding constraints are connectivity (about 900 million Africans offline), energy and compute (0.6% of global data centre capacity), language coverage (88% of African languages poorly served by language technology) and weak governance capacity. The paper closes with recommendations for governments, firms, development partners and researchers: set poverty targets in AI strategies, treat connectivity and compute as AI policy, build African-language data, protect workers and users, and fund outcome evaluations.
James Mane · 2026 · Africa