As education increasingly shifts towards technology-driven models, artificial intelligence systems like GPT are gaining prominence for their potential to enhance educational support. In both university and MOOC environments, students often face the challenge of selecting courses that align with their individual academic needs. Providing access to detailed information about the knowledge concepts covered in each course can facilitate more informed decision-making, but manually curating this information is labor-intensive and time-consuming. This chapter investigates the ability of GPT to generate relevant knowledge concepts from course syllabi. These AI-generated concepts are then used to construct a comprehensive knowledge graph, which is integrated into an educational recommendation system. We evaluate the quality of the GPT-generated concepts at both the concept and course levels, comparing them to existing educational resources. The results indicate that the quality of the GPT-generated concepts is high and consistent with course content. Furthermore, applying the knowledge graph built from these GPT-generated concepts to the recommendation system enhances its accuracy and performance. Our findings demonstrate the significant potential of GPT-generated knowledge graphs in educational technology. By providing a more efficient and scalable method for personalized course recommendations, this approach offers the possibility of greatly enhancing the learning experience.