A Data-Driven and Explainable AI Model for Human-Centric Artificial Intelligence Readiness Among Primary Teachers

Abstract

As Artificial Intelligence (AI) integration becomes a priority in global K-12 education, understanding teacher readiness is important. Traditional discourse often adopts a technology-centric approach, frequently overlooking the complexity of human factors. This study develops a human-centric predictive model to evaluate AI readiness through digital literacy, self-efficacy, and infrastructure access, using 217 valid responses from Jakarta, Banten, and West Java. Through cross-model validation combining Multiple Linear Regression (MLR) and Classification and Regression Tree (CRT), this study evaluated the hierarchy of influence between variables. The framework explains 60.7% of the cross-validated variance (78.8% on testing data), with self-efficacy as the dominant predictor. The findings indicates that internal psychological factors have a notably higher predictive weight than structural infrastructure, suggesting that sustainable AI adoption must prioritize psychological intervention over hardware investments.