Generative AI as a Teaching Partner for Personalised Learning Paths — Dar es Salaam and Dodoma, Tanzania

Researcher Juliana Kamaghe of the Open University of Tanzania studied how generative AI tools can help secondary-school teachers build personalised learning paths for students, across schools in Dar es Salaam (urban) and Dodoma (semi-urban/rural).

120 teachers used accessible generative AI tools - principally ChatGPT and Grok, alongside Curipod and MagicSchool.ai - to generate weekly lesson plans aligned to national syllabus objectives, automatically produce quizzes with adaptive difficulty, and tailor content to student diagnostic profiles. The study used a convergent parallel mixed-methods design: a quantitative pre/post strand (paired t-tests) on student engagement, test performance and teacher workload, combined with usage logs, open-ended surveys and focus groups.

Reported results include student engagement rising from 2.8 to 4.3 on a 5-point scale and mean test scores rising from 61% to 75%, alongside teacher-reported reductions in lesson-prep workload; teachers described the tools as intuitive and useful for differentiating instruction. There was no no-AI control group, so the reported gains cannot be cleanly separated from novelty effects, teacher enthusiasm, or concurrent instruction changes, and self-reported workload/engagement measures may be optimistic.

The authors are candid that rural schools, more prevalent around Dodoma, face persistent connectivity, device and training gaps that limit equitable scale-up, positioning this as an encouraging pilot rather than a proven, ready-to-scale model.