<![CDATA[Purpose: The rapid advancement of artificial intelligence (AI), particularly deep learning, has created new opportunities for enhancing learning in mathematics and physics education. This study aims to develop and modify a deep learning based chatbot model to improve students’ conceptual understanding of advanced mathematical concepts, including calculus, linear algebra, and differential equations, which are widely applied in physics-related learning at the undergraduate level. Method: This study employed a research and development (R&D) approach consisting of needs analysis, model design, implementation, and evaluation stages. The chatbot was developed by integrating deep learning algorithms with natural language processing (NLP) and domain-specific mathematical training datasets. The implementation involved undergraduate students at STKIP Muhammadiyah Sungai Penuh. Data were collected through pre-tests and post-tests to measure conceptual understanding, student questionnaires to examine learning motivation and usability, and expert validation to assess content validity and pedagogical suitability. Quantitative data were analyzed using normalized gain scores and paired statistical testing, while qualitative feedback was examined descriptively. Findings: The results show that the modified chatbot significantly improved students’ conceptual understanding of advanced mathematical concepts compared to conventional learning approaches. Students demonstrated higher learning gains and increased motivation, particularly due to the chatbot’s ability to provide contextual feedback, step-by-step problem-solving assistance, and adaptive learning support. Expert evaluations confirmed the practicality and educational relevance of the developed chatbot. Significance: This study demonstrates that AI-powered chatbots can function as effective supplementary learning tools in higher education mathematics. By bridging advanced mathematical content with physics-related applications, the chatbot supports independent learning and contributes to the development of intelligent educational technologies for complex STEM subjects.]]>
