This paper explores the utilization of large language models (LLMs) in introductory programming courses in higher education. The primary objective is to examine how LLMs can support the creation of programming-related teaching materials. In addition, the aim is to reflect their potential impact on the learning objectives of these courses. The study is conducted as a literature review. The findings indicate that LLMs can be employed to generate a variety of programming tasks, such as small programming exercises and multiple-choice questions, with the quality of these tasks often comparable to those created by humans. Furthermore, LLMs enable the development of novel assignment types. In terms of learning objectives, the increasing prevalence of LLMs highlights the potential to introduce more advanced course content earlier. If LLMs are used to generate code, instructional emphasis may shift from code writing to reading, analyzing, and modifying code. Nevertheless, teaching fundamental programming concepts seems to remain essential, necessitating a balanced approach between leveraging LLMs and maintaining traditional programming practices.
