Artificial Intelligence-Based Teaching Support System for Personalized Learning in Higher Education

Abstract

The aim of this study was to design and evaluate an artificial intelligence-driven teacher support system capable of translating multidimensional learning profiles into instructional and assessment recommendations for higher education. The proposal integrated five diagnostic dimensions: VARK, for preferences in accessing information; FSLSM, for processing and representation traits; ICAP, for levels of cognitive engagement; SRL, for self-regulated learning; and EBSI, for study habits. An explanatory sequential mixed-methods approach was used. In the quantitative phase, a digital questionnaire was administered to 145 students from five universities, and the data were processed through a system developed in Python, which generated diagnostic profiles and personalized pedagogical reports for teachers. In the qualitative phase, four instructors evaluated the usefulness, perceived coherence, feasibility, and ethical acceptability of the reports. The results showed substantial convergence between the diagnosis calculated without AI and the narrative generated with AI in dimensions such as VARK and ICAP, as well as favorable acceptance by the teaching staff. However, mismatches were identified in the calibration of FSLSM and in the intensity attributed to SRL and EBSI. It is concluded that AI can mediate instructional decision-making under pedagogical supervision