Uso de un tutor virtual basado en inteligencia artificial como apoyo al aprendizaje del álgebra elemental en Octavo EGB de la Unidad Educativa La Providencia

Abstract

Abstract: This study analyzed the changes observed in conceptual errors and performance in elementary algebra among 23 eighth-grade students from the La Providencia Educational Unit, following the implementation of a didactic strategy mediated by the AI Chat module of the AI Didactic Platform. The study employed a quantitative approach, a positivist paradigm, and a single-group pre-experimental design with a pretest and posttest. Educational assessment was used as the technique, consisting of a structured eight-item test covering monomials, coefficients, variables, literal parts, like terms, and the simplification of expressions. The mean score increased from 7.00 on the pretest to 8.87 on the posttest. As the differences did not meet the assumption of normality (Shapiro-Wilk = 0.762; p < 0.001), the Wilcoxon signed-rank test was applied, yielding W = 0 and p < 0.001. A descriptive reduction in conceptual errors was also observed. Due to the single-group design lacking a control group and the presence of threats to internal validity, these changes cannot be attributed exclusively to the virtual tutor.