Adaptiv'World (EvidenceB) — AI-Adaptive Learning Lifts Math and French Scores at a TVET School in Gagnoa, Côte d'Ivoire

Under the World Bank's Youth-RISE project, funded by the Mastercard Foundation and run with Côte d'Ivoire's Ministry of Technical and Vocational Education and Training (METFPA), around 2,000 students across 25 TVET institutions — including the Technical and Vocational High School of Gagnoa — were given access to the EvidenceB adaptive-learning platform for maths and French over a 22-week period in 2024-2025.

The platform pairs a multi-armed-bandit algorithm that selects exercises matched to each student's demonstrated level with a clustering algorithm that groups students by learning characteristics. Of the roughly 2,000 targeted students, 1,529 (76.5%) actively engaged with the platform.

Using a difference-in-differences design against non-user comparison groups, the World Bank's evaluation found active users gained 0.234 standard deviations in maths (p<0.01) and 0.121 SD in French (p<0.05) — equivalent to roughly 11.2 and 5.8 additional months of learning. Students in the bottom 15% of the baseline performance distribution progressed 6 to 53 times faster than higher-performing peers.

The evaluation is candid about what remains unresolved: digital infrastructure needs strengthening, curriculum-alignment questions are unsettled, and no teacher-training model for large-scale rollout has yet been developed, so the gains recorded so far rest on a contained pilot rather than a proven system-wide deployment.