Tailoring STEM Education: A Comparative Analysis of GPT-4 Omni and Gemini Advanced for Personalized Learning Experiences

Abstract: The need for personalized learning in Science, Technology, Engineering, and Mathematics (STEM) education is increasingly recognized as a means to address diverse learning needs and improve student outcomes. This study presents a comparative analysis of two Large Language Models (LLMs), GPT-4 Omni and Gemini Advanced, to assess their effectiveness in enhancing personalized learning experiences for STEM students. The study evaluated both tools for their abilities to adapt to diverse learning styles and their impact on student engagement, learning outcomes and motivation. Using a mixed-methods approach that combined quantitative and qualitative data, the study found that both LLMs offer unique capabilities for personalization, with GPT-4 Omni showing strengths in providing tailored content creation, and Gemini Advanced demonstrating a more adaptive feedback mechanism. However, both tools also presented some practical challenges, including the need for continuous adaptation to individual student needs. The findings have important implications for educators and educational technology developers, highlighting both the potential and the limitations of integrating LLMs into personalized learning environments. Educators are advised to use LLMs strategically, focusing on specific personalization goals, and with robust oversight. Educational technology developers should prioritize the creation of platforms that facilitate effective teacher involvement, and address the identified ethical and bias concerns, to ensure that these tools have a positive impact on personalized learning in STEM education. Keywords: AI, Research, GPT-4, Gemini Advanced