With the continuous progress of artificial intelligence algorithms, music creation has gradually developed from purely manual operation to intelligent assistance, a process that not only optimizes the efficiency of music creation, but also opens up new paths of creative thinking. This study explores the impact of artificial intelligence on creative thinking in music composition and teaching. By introducing AI-based music composition models and teaching methods, its performance in practical applications was analyzed. The study used scoring data from 30 professional listeners and 25 general listeners and compared them with the AI scoring system. The results show that the difference between the AI ratings and the ratings of the professional listeners is about 1.5 points, indicating that AI can more accurately simulate the perception of professional listeners in music composition evaluation. In addition, the model training generated the highest music scores at an iteration number of 60k, which were 79.249 points for the artificial rating and 80 points for the model rating, respectively. The study shows that AI has a significant auxiliary role in music creation and teaching, which can promote the diversity of music creation and personalization of teaching, and enhance students’ understanding and interest in music.