Computer Programming Education in the Age of Generative AI: Insights from Empirical Research

The integration of Generative Artificial Intelligence (GenAI) tools in programming education has become a focal point in recent years. This research delves into existing literature to explore the utilization of GenAI tools for teaching programming across different educational levels. Through a systematic review spanning from 2018 to 2023, we aim to offer insights into the current status of GenAI in programming education. Our analysis of 37 relevant studies indicates a predominant emphasis on higher education, constituting 81% of the reviewed literature. Among the findings, we identify six programming language instructions leveraging GenAI tools, with Python being the most prominent, featured in nearly half (49%) of the studies. The study also pinpoints primary GenAI applications such as Google's Bard (now Gemini), Codex, and Copilot, with ChatGPT emerging as the most commonly utilized tool for programming education. Objectives of the reviewed studies range from evaluating the performance of Large Language Models (LLMs) to understanding student and teacher perceptions regarding their usage. Additionally, these objectives encompass providing code explanation and suggestions, error correction, feedback provision, and integration with supportive approaches. Moreover, our analysis outlines significant learning outcomes associated with the adoption of GenAI in programming education, including increased motivation, interest, and engagement among students. Despite the potential benefits, the study identifies challenges faced by educators and students alike. Approximately 22% of the reviewed studies highlight various difficulties and concerns, ranging from issues with grading feedback to worries about student over-reliance on GenAI tools, which may lead to incorrect solutions and hinder learning progress. Overall, our study contributes to a comprehensive understanding of how GenAI tools are utilized in programming education. By analyzing the objectives and findings of the reviewed studies, we shed light on the evolving landscape of GenAI adoption and its implications for programming education.