A systematic review of research on large language models for computer programming education

Given the increasing demands in computer programming education and the rapid advancement of large language models (LLMs), LLMs play a critical role in programming education. This study provides a systematic review of selected empirical studies on LLMs in computer programming education, published from 2023 to March 2024. The data for this review were collected from Web of Science (SCI/SSCI), SCOPUS, and EBSCOhost databases, as well as three conference proceedings specialized in computer programming education. In total, 42 studies met the selection criteria and were reviewed using methods, including bibliometric analysis, thematic analysis, and structural topic modeling. This study offers an overview of the current state of LLMs in computer programming education research. It outlines LLMs' applications, benefits, limitations, concerns, and implications for future research and practices, establishing connections between LLMs and their practical use in computer programming education. This review also provides examples and valuable insights for instructional designers, instructors, and learners. Additionally, a conceptual framework is proposed to guide education practitioners in integrating LLMs into computer programming education. This study suggests future research directions from various perspectives, emphasizing the need to expand research methods and topics in computer programming education as LLMs evolve. Additionally, future research in the field should incorporate collaborative, interdisciplinary, and transdisciplinary efforts on a large scale, focusing on longitudinal research and development initiatives.