AI in Education
AI tutoring, adaptive paths, and automated feedback designed for African classrooms — not retrofitted from Western consumer tools.
- AI tutoring
- Adaptive learning paths
- Automated feedback on student work
- Personalized study schedules
- Teacher-facing AI assistants
Why an African lens matters
AI tutoring is a fast-moving global field, but most consumer products are tuned for English-first learners with broadband access and a tutor’s worth of one-on-one screen time. African schools rarely have any of those conditions. A useful AI tutor here is one that:
- Handles low-bandwidth and intermittent connectivity gracefully
- Speaks the languages students think in — Amharic, Afaan Oromo, Tigrinya — not just English
- Knows the local curriculum and exam frame, not just generic textbook material
- Augments teachers rather than replacing them, because there are not enough teachers to begin with
That last point matters most. The opportunity is not to build a tutor that replaces a teacher; it is to build tools that let one teacher reach thirty more students well.
What we’re researching
- AI tutoring. Conversational tutors grounded in the Ethiopian curriculum, with explicit citation back to textbook sources so students can verify what the model says.
- Adaptive learning paths. Sequencing problems by difficulty + topic mastery, not a fixed chapter order — so a struggling student gets more practice on prerequisites, and an advanced one moves ahead.
- Automated feedback on student work. Short-answer and essay feedback in Amharic and English. Calibrated against teacher-graded baselines.
- Personalized study schedules. Spaced-repetition scheduling around the ESSLCE and university entrance exam dates, with realistic time budgets for students who also work or farm.
- Teacher-facing AI assistants. Lesson planning, question generation, and grading aids so teachers spend less time on paperwork and more on teaching.
Where we are
This pillar is prototyping. We have working internal experiments with retrieval-grounded tutoring against the Ethiopian secondary curriculum and an automated short-answer grader in early evaluation. No public release yet — accuracy floors and safety guarantees are still being established.
If you are a researcher working on multilingual education AI or an Ethiopian school willing to pilot a teacher-facing tool, reach out.