How might low- and middle-income countries create practical ways to use AI to enhance learning outcomes, so that today’s standards are achieved by more learners?
EdTech Hub’s AI Observatory is exploring how education systems can renew the purpose of learning – which means rethinking what education is for in a world shaped by AI, with the aim of narrowing the learning divide.
This week, in Issue No. 31 of the #WaypointWednesday, we spotlight ways AI is being used to accelerate reading assessment, improve participation from learners with diverse needs, and reduce the admin burden on teachers to focus on the relational aspects of education.
Early signals
AI used to accelerate reading assessment
A major challenge in LMICs is large student populations combined with time constraints, which can result in superficial grading and inconsistent standards (Yakubu, 2024). We’re seeing AI used to streamline assessment processes.
- India – Oral Reading Fluency (ORF):The ORF tool listens to students read aloud and automatically transcribes their speech. It significantly accelerates the assessment process and helps capture digital reading records for each learner. As of March 2025, the tool has been used to conduct over 3.6 million oral-read assessments across 33,000 schools in Gujarat, in partnership with the state government. (Wadhwani AI, 2025; Adam & Lester, 2025)
- South Africa – EGRA-AI: Automates and scales foundational reading assessments in Sepedi and isiXhosa. This innovation aims to reduce the costs and resource demands of traditional one-on-one evaluations while improving accessibility and accuracy. The AI model for isiXhosa achieved ~95% item-level accuracy for items where all three human annotators agreed. Technical issues included background noise, difficulty deciphering short utterances, and lower performance with letters than with words. (AI-for-Education, 2025; Adam & Lester, 2025)
AI used to expand access to learners with disabilities
We’re seeing AI used to widen early identification, reduce bottlenecks in specialist services, and improve participation for learners with diverse needs.
- Philippines – ECAIR’s SABAY:Philippines ECAIR (Education Center for AI Research) is using AI to tackle the hidden costs of screening in education, where speech-language disorders and dyslexia may otherwise go undetected and untreated, even after preliminary tests. The SABAY AI-assisted system acts as the first line of defense, used to filter learners who need specialists versus those who don’t. (DepEd Philippines, 2026)
- Kenya – Artificial Intelligence-technology for Kenyan Sign Language: Researchers at Maseno University, Kenya, developed assistive AI technology that converts spoken English to Kenyan Sign Language (KSL). The initiative aims to break communication barriers between deaf students and their peers and teachers in mainstream schools, increase learning outputs, and sustain inclusion in schools. (Ayere et al., 2024; Adam & Lester, 2026)
AI used to free up teachers to teach
While evidence is still emerging on whether AI genuinely saves teachers’ time or creates new hidden forms of work (Adam & Lester, 2025), we’re seeing AI framed as a tool to free up teachers to focus more on the pedagogical and relational aspects of their work.
- Pakistan – low-resource government schools: A case study from low-resource government schools in Pakistan suggests that scalable, mobile-based AI training coupled with low-cost, context-sensitive AI interventions can enhance teaching quality. After the intervention, 98% of teachers integrated AI into daily teaching, and approximately 70% reported improved lesson delivery and reduced workload. (Hafeez & Zehra, 2025; Adam & Lester, 2026)
- Rwanda – Tari: An AI-powered teaching assistant that is being piloted in Rwanda and is one of EdTech Hub’s Teachers-in-the-Lead sandboxes; results are expected in early 2026 (EdTech Hub, 2025; Rising Academies, 2025). It supports subject knowledge assistance, lesson planning, generating instructional materials and activities, and supporting student assessments. (Adam & Lester, 2026)
Reflections
- Studies are emerging illustrating that AI can support autistic learners with differentiated learning opportunities, such as through conversational practice with ChatGPT and supporting emotional regulation. However, caution should be taken regarding sensory and information overload. (Adam & Lester, 2026)
- While AI has great potential to improve learning outcomes if used in appropriate ways, there are still broader concerns such as misuse of student data, breaches, misleading feedback and discrimination due to biased model training. (Stefanus et al., 2025; Ogbu Eke, 2024)
Swipe for a quick take
We’d love to hear from you! What’s been shaping your thinking on AI? Drop your thoughts (and reading recommendations) in the comments. Explore more from EdTech Hub’s AI Observatory.
EdTech Hub’s AI Observatory is made possible with the support of the UK’s Foreign, Commonwealth and Development Office.
