Systematic review of the implementation of LLMs as teaching-learning tutoring tools in the educational field

Abstract

This Systematic Literature Review (SLR) examines the effectiveness of Large Language Models (LLMs) as educational tutoring tools. The study analyzes their impact on learning outcomes and academic performance when compared to conventional pedagogical methods. LLMs offer key empirical benefits such as real-time feedback, educational personalization, scalability in hybrid environments, and significant interdisciplinary potential, positioning them as essential tools for educational modernization. The primary objective was to systematically analyze the efficacy of LLMs as intelligent tutors, identifying differences in academic performance, ethical challenges, implementation strategies, and motivational factors. A systematic search was conducted in Scopus using the PICO-PRISMA framework. From a total of 973 initial records, 28 open-access studies published in English (between 2022 and 2025) were selected. The evidence gathered reveals significant improvements in academic performance, highlights critical ethical challenges, and presents structured pedagogical strategies with an 86.4% success rate. This analysis provides consolidated and valuable information for future developments and implementations of LLMs in the educational field.