Exploring Intelligent Tutoring Systems: Integrating GAI and Advanced Pedagogy for Design, Challenges, and Potentia

Abstract

This article addresses the experience achieved in the development of an Intelligent Tutor System (STI), which has been the result of a continuous research process that the research team has been carrying out for educational innovation in the field of engineering. The STI itself is presented, revealing its architecture and its main functional modules, as well as the incorporation of pedagogical methodologies to provide a teaching structure and adaptability such as those established by Bloom's Taxonomy and the Learning Styles of the Vark model. In addition, how Information and Communication Technologies (ICTs) have been integrated is addressed, among which are Artificial Generative Intelligence (IAG), Natural Language Processing (NLP), Large Language Models (LLM), and Model-View-View Model (MVVM). The pedagogical methodologies and ICTs indicated above have been put into operation to complement the process of Knowledge Space, Knowledge Structure, Knowledge Construct, and Deep Knowledge Tracing, to have an ITS that has the capabilities of directing an educational process, equalizing or outperforming the average human teacher. This modular system and service-oriented architecture take advantage of the Applications Programs Interfaces (APIs) of advanced models such as the Generative Pre-trained Transformed (GPT) to generate personalized educational content, offering a scalable and efficient platform. The use of this STI allows to offer quality education for engineering areas, both in terms of pedagogy and technical aspects. With this project, we seek to put into operation a disruptive educational platform that changes the teaching-learning process, the way it is being carried out, and the learning experience by the students. At the same time, it can also become a reference framework for the design and development of STIs for those researchers who wish to start creating one