Implementación de un prototipo de un sistema para apoyar tareas de enseñanza-aprendizaje: Un enfoque basado en Agentic AI y modelos generativos

Abstract

Abstract: This paper develops and implements a prototype system based on Agentic AI to support teaching and learning tasks in higher education, with the aim of promoting the use of intelligent systems in this field. The Agentic AI prototype seeks to better leverage the generative capabilities of large language models to obtain smarter and more contextualized educational tools that assist those involved in the teaching and learning process. The system was developed using a modular architecture that employs traceable agents and tools, which enable the generation of contextualized responses for students through information retrieval and web searches, as well as the creation of structured lesson plans based on the ADDIE model and the analysis of real-world needs. The evaluation considered quantitative, qualitative, and performance criteria, as well as a comparison between language models. The results obtained confirm the technical feasibility of the approach in a controlled environment and open the possibility of further pedagogical validation and institutional application.