WASSCElab: Chatbot integration in an exam preparation platform

WASSCElab is a digital learning platform providing access to past exam questions and structured resources aligned with the West Africa Examinations Council (WAEC) syllabus. With our support, Center for Integrated Solutions (CIS) integrated an GPT-4o-powered student chatbot into the platform to offer additional support for exam preparation in Liberia.

Why this is important
Exam results in the widely used WAEC exams fluctuate significantly, especially in Liberia. Students struggle with basic literacy and access to learning resources for exam prep. An AI support chatbot could address these issues and inform learning on developing individualised and well contextualised student support tools.

Key Learnings

1. Co-designing with teachers and students is non-negotiable - Teacher resistance to AI tools often comes from unfamiliar methods, lack of ownership or fear of loss of control. Involving teachers early can build trust and long-term adoption. User testing also suggests student engagement must be central to AI design.
2. Early model evaluation is key before scaling - Quality assurance testing by subject matter experts with reference questions was important to reveal model limitations. Foundational models, largely trained on broad, non-local content, struggled with context-heavy subjects (History, Geography, Literature).
3. Context matters - Building larger, more detailed, and context-specific data sources is needed to fully realise the potential of AI in countries like Liberia.

Grant we provided: $42,320