AI IN EDUCATION DISPATCH | 5 MIN READ

**Five stories on the gap between committing to AI in education and being able to afford it, staff it, or govern it**

Two World Bank reports arrive within a month of each other. One measures what AI tutoring actually delivers in low- and middle-income classrooms. The other asks what it would cost a ministry to run. Read together, they are the clearest statement yet of the distance between the promise and the budget line.

1 | GLOBAL | WORLD BANK

World Development Report 2026 finds AI tutoring delivers real learning gains in LMICs, and that capturing them depends on keeping teachers central, adapting tools to local languages and curricula, and building a far stronger evidence base

The WDR 2026 devotes significant attention to the implications of AI for education as part of its broader analysis of AI and development. The report presents AI as offering real promise for improving teaching and learning, particularly in LMICs. A synthesis of randomised experiments across China, Ghana, Nigeria, Türkiye and the United States finds AI-powered tutoring produces meaningful gains, and in Ghana, maths tutoring delivered by text message on basic phones generated nearly a full year of learning at about five dollars per student.
But the report’s key message is adopt, adapt, advance. Rather than viewing AI as a technology project, it frames it as an implementation challenge. The biggest gains are expected to come from integrating AI into existing education systems: supporting teachers, fitting local curricula and languages, and strengthening the institutional conditions needed for sustained use.
Instead of asking how to deploy the latest AI models, systems should identify where AI can relieve known bottlenecks such as shortages of teacher expertise, lesson preparation or individualised student support, then introduce proven tools that fit local conditions. Only after building experience, evidence and institutional capability should countries move toward developing more sophisticated AI solutions of their own.
READ MORE AT WORLD BANK | AUGUST 2026

2 | GLOBAL | WORLD BANK

A new World Bank working paper asks how ministries of education can actually afford AI under real fiscal and infrastructure constraints, and answers with six approaches that begin with teachers rather than students

Costing AI Use in Education in Low- and Middle-Income Countries explores how ministries of education can sustainably finance AI for teaching and learning. Given fiscal and infrastructure constraints, it recommends starting with affordable, scalable approaches that limit initial investment, contain recurring costs, and generate evidence to inform policy and future investment.
Rather than starting with large-scale, systemwide deployment, countries can begin with targeted applications that address binding constraints to teaching and learning while building institutional capacity.
The six approaches it identifies: start with teachers, not students; weigh AI costs against the high cost of the status quo; build on infrastructure already in place; rethink financing and procurement; design for low connectivity and limited budgets; and treat evidence generation as a core investment rather than an afterthought.
READ MORE AT WORLD BANK | JULY 2026

3 | INDIA | PUNJAB

Punjab becomes the first Indian state to teach AI in every class from 1 to 12, reaching 3.15 million students across 25,000 schools by training 200,000 teachers to deliver it

The rollout, named Sarbat.AI, comes in two phases. AI concepts fold into the existing Computer Science curriculum for Classes 8 to 12 first, then extend to Classes 1 to 7 through age-appropriate content built to introduce foundational digital literacy.
The teacher figures are the more telling part. 12,400 Computer Science teachers are being trained in the first phase, with the programme eventually reaching more than 200,000 teachers across other disciplines so AI can be taught inside existing subjects rather than alongside them.
The curriculum was developed after examining models in Singapore, Finland, the UAE and the United States, and is aligned to UNESCO’s Ethical AI Framework. Neeev AI leads implementation, with Google, Intel, Canva and FA-AI contributing literacy programmes, teacher training and innovation support. The statewide launch follows a state board pilot that ended in a student AI hackathon.
READ MORE AT ELETS DIGITALLEARNING | 27 JULY 2026

4 | MALAWI | EDTECH HUB

Fewer than one in three Malawian children with disabilities attends school, and EdTech Hub argues technology can close part of the gap left by a national inclusive education strategy that expired in 2021 and was never replaced

An estimated 14% of children aged 5 to 17 in Malawi have at least one functional difficulty. The country holds a strong rights-based legal framework and has signed the international commitments, but the lapse of the National Inclusive Education Strategy has left teachers and education personnel without unified national guidance on identifying, assessing and supporting learners with special educational needs and disabilities.
Acute specialist teacher shortages, limited data disaggregation and fragmented resource allocation compound the problem.
The policy brief makes four recommendations: finalise a tech-forward national inclusive education policy, modernise data systems for more nuanced SEND identification, use digital platforms for scalable teacher capacity building, and bridge the innovation-to-scale gap for EdTech and assistive devices.
READ MORE AT EDTECH HUB | 1 AUGUST 2026

5 | EAST AFRICA | IUCEA

East Africa’s universities launched a shared AI network in Nairobi to pool research and infrastructure across the East African higher education community after years of national efforts running in parallel rather than together

The Inter-University Council for East Africa has launched a Regional AI Network of Universities and a new Strategic Plan for 2026 to 2031 to strengthen AI education, research and innovation across the East African Community. The three-day conference, convened with Kenya’s Ministry of Education, drew more than 400 participants.
The initiative aims to move beyond fragmented national efforts by connecting universities, pooling expertise and infrastructure, and promoting collaboration on responsible AI. It extends work IUCEA began in Kigali in March under the EAC AI Alliance, framed then as a response to national AI efforts that have limited the scale and impact of investment across the eight EAC partner states.
What distinguishes the approach is that it operates at the regional-bloc level rather than the institutional or national one, with a mandate covering quality assurance, academic mobility and qualifications recognition. Whether that translates into pooled compute, shared datasets and genuinely joint research programmes, rather than a coordinating layer above institutions that carry on working separately, is what the plan will be judged on.
READ MORE AT IAFRICA | 3 AUGUST 2026

Every source in this issue was checked against its published page metadata before inclusion.

8. The Implementation Gap: What It Actually Takes to Afford, Staff, and Govern AI in Education

Five stories on the gap between committing to AI in education and being able to afford it, staff it, or govern it

World Development Report 2026 finds AI tutoring delivers real learning gains in LMICs, and that capturing them depends on keeping teachers central, adapting tools to local languages and curricula, and building a far stronger evidence base

A new World Bank working paper asks how ministries of education can actually afford AI under real fiscal and infrastructure constraints, and answers with six approaches that begin with teachers rather than students

Punjab becomes the first Indian state to teach AI in every class from 1 to 12, reaching 3.15 million students across 25,000 schools by training 200,000 teachers to deliver it

Fewer than one in three Malawian children with disabilities attends school, and EdTech Hub argues technology can close part of the gap left by a national inclusive education strategy that expired in 2021 and was never replaced

East Africa’s universities launched a shared AI network in Nairobi to pool research and infrastructure across the East African higher education community after years of national efforts running in parallel rather than together

Leave us your email to stay informed

8. The Implementation Gap: What It Actually Takes to Afford, Staff, and Govern AI in Education

Five stories on the gap between committing to AI in education and being able to afford it, staff it, or govern it

World Development Report 2026 finds AI tutoring delivers real learning gains in LMICs, and that capturing them depends on keeping teachers central, adapting tools to local languages and curricula, and building a far stronger evidence base

A new World Bank working paper asks how ministries of education can actually afford AI under real fiscal and infrastructure constraints, and answers with six approaches that begin with teachers rather than students

Punjab becomes the first Indian state to teach AI in every class from 1 to 12, reaching 3.15 million students across 25,000 schools by training 200,000 teachers to deliver it

Fewer than one in three Malawian children with disabilities attends school, and EdTech Hub argues technology can close part of the gap left by a national inclusive education strategy that expired in 2021 and was never replaced

East Africa’s universities launched a shared AI network in Nairobi to pool research and infrastructure across the East African higher education community after years of national efforts running in parallel rather than together

7. The Reading Bet: Why AI's Clearest Wins in Education Are Showing Up in Literacy

Leave us your email to stay informed