This study explored the potential of generative artificial intelligence–based simulation construction activities to facilitate changes in students’ academic achievement and STEAM competencies in chemistry. The study was conducted with 54 second-year students enrolled in an advanced chemistry course at a science high school in Gyeongsangbuk-do, Republic of Korea. With the support of ChatGPT, students constructed particle-level simulations based on VPython and engaged in inquiry activities involving iterative modification and refinement of their simulations. Academic achievement was measured using five items related to acidbase equilibrium administered before and after the intervention, and STEAM competencies were assessed using a 21-item, four-point Likert-scale questionnaire. The results indicated that academic achievement scores increased from pretest to posttest. Differences across achievement levels were also observed, with students in the higher-achievement group showing greater gains. STEAM competencies also showed an overall increase in mean scores, and the largest change was observed in creative problem-solving. These findings suggest that the use of generative artificial intelligence in chemistry instruction may be related to context-specific changes in students’ conceptual understanding and competencies, and provide implications for future research in generative artificial intelligence–supported chemistry education.