Indonesia AI-Powered Computer-Based Assessment for Literacy and Numeracy: Multi-School Validation in Indonesian Secondary Education

A Hindriana, Rio Priantama, Ina Setiawati, A R PratamiNon-Scopus1 Januari 2026Afiliasi: ID

Abstrak

Accurately diagnosing literacy and numeracy competencies remains a persistent challenge in developing countries despite significant investments in educational technology. This study developed and validated Asesmen Kompetensi Minimum (AKM) Online, an Artificial Intelligence (AI)-enhanced Computer-Based Assessment (CBA) system that integrates Fisher–Yates Shuffle and Regular Expression algorithms to enable secure test randomization, automated response validation, and real-time diagnostic feedback at scale. Through multi-school implementation across three contrasting socioeconomic contexts (N = 552), the system generated 17,049 student responses and demonstrated high technical reliability (AI scoring accuracy = 94.35%). Results revealed severe and consistent numeracy deficiencies (95.8–100% requiring intervention) and substantial variation in literacy needs (53.9–78.46%), indicating that contextual factors influence literacy more strongly than numeracy. Mathematical reasoning showed complete failure (0% proficiency), a finding corroborated through expert validation, cross-context replication, and alignment with national and international benchmarks). Expert evaluations confirmed high system quality in both educational assessment (81%) and informatics performance (82%). The study provides empirical evidence that AI-supported CBA can overcome major limitations of existing national assessments by enabling whole-population testing with immediate, actionable competence profiles. The scalability, transparency, and resource-efficient design of AKM Online underscore its potential for broader adoption in developing countries seeking reliable, evidence-based approaches to foundational literacy and numeracy assessment.

Identifier

Ekspor BibTeX
Scroll to Top