Personalized Mathematics Teaching Supported by AI Chatbots to Enhance University Students’ Mathematical Problem-Solving Competence in Vietnam

This study examines the effectiveness of personalized mathematics teaching supported by AI chatbots in enhancing university students’ mathematical problem-solving competence in Vietnam. Focusing on first- and second-year undergraduates enrolled in "Applied Statistics" and "History of Mathematics" courses, we conducted a mixed-methods quasi-experimental study (N = 200) to evaluate the impact of AI chatbot-supported prompt-engineering training on problem-solving competence. Using a quasi-experimental pre-test/post-test design with control and experimental groups, we measured learning gains through adapted PISA/TIMSS-based assessments and examined relationships among prompt-engineering skills, AI-related attitudes, and problem-solving performance using regression analysis. Quantitative results showed that the experimental group achieved higher post-test problem-solving scores than the control group, with a large standardized mean difference (Cohen’s d = 1.36). After controlling for pre-test scores, the ANCOVA also showed a statistically significant group effect F(1, 197) = 91.76. Qualitative interviews with 20 students suggested that structured prompting helped students improve confidence, clarify mathematical concepts, and recognize the need to critically evaluate chatbot-generated responses. The findings provide initial empirical support for structured prompt-engineering training in AI chatbot-assisted mathematics learning in higher education. The study offers practical implications for integrating prompt-engineering instruction into AI-supported mathematics teaching and contributes to research on technology-enhanced learning in Vietnamese higher education.