Generative AI for Programming Education: Can ChatGPT Facilitate the Acquisition of Fundamental Programming Skills for Novices?

Abstract

Modern Generative AI (GAI) systems like ChatGPT have sparked much interest in their potential to revolutionize programming education, especially for beginners. However, the existing empirical data regarding the effectiveness of technologies like ChatGPT as autonomous programming tutors is presently limited. The present study investigates the capacity of ChatGPT to facilitate the acquisition of fundamental programming skills for novice programmers without human assistance. This study puts forth a conceptual framework (APEC - Adaptive Programming Education via ChatGPT) that integrates both bottom-up and top-down approaches, incorporating ChatGPT as the principal instructor for the study of programming. An empirical study was undertaken to assess the usefulness of ChatGPT as a tool for teaching novice programmers a new programming language. This empirical study was conducted on 20 undergraduate students. To provide an expert assessment of the quality of the responses, a survey was conducted with three programming experts proficient in Python. The survey findings indicate that ChatGPT is proficient in explaining core principles such as variables, data types, and control statements through conversational exchanges, adopting an intelligent and logical methodology. Nevertheless, certain constraints arise when dealing with increasingly complex topics.