A team of researchers from ETH Zurich, Charité – Universitätsmedizin Berlin, and Kwame AI Inc have developed and piloted "Adesua," a WhatsApp-based AI teaching assistant delivering personalised science education to students in West Africa.
According to the teams, the platform builds upon Kwame for Science, upgrading its architecture to handle direct conversational question-answering and comprehensive automated test assessments.
Kwame for Science is an AI-powered teaching assistant designed to support science education for students across West Africa. Named after Ghana’s first president, Dr Kwame Nkrumah, the platform provides instant answers to curriculum-based science questions.
The researchers state that the primary innovation driving Adesua is its localised, high-throughput Retrieval-Augmented Generation pipeline, which mitigates large language model hallucinations.
They anchored a GPT-4 architecture within a curated knowledge base featuring local textbooks and 33 years of national examination questions from the Basic Education Certificate Examination and the West African Senior School Certificate Examination.
Adesua's operational feasibility was evaluated during a six-month pilot deployment in 2025 involving 56 active users in Ghana.
The platform features an interactive assessment system allowing learners to take timed or untimed multiple-choice quizzes, automatically grading inputs to generate detailed performance reports delivered as PDFs via WhatsApp.
Preliminary logging revealed a high perceived utility, yielding a 93.75% helpfulness score.
Adesua enters an emerging competitive landscape of conversational mobile learning tools addressing Africa’s severe shortage of qualified educators.
Share
