Artificial intelligence in coding education: a study on self-efficacy, attitude, and achievement

The aim of the study is to compare artificial intelligence supported coding education, delivered using different chatbots within a specific teaching plan, with traditional compiler-based instruction in terms of science high‑school students’ programming self-efficacy, attitude towards programming and academic achievement. A quasi-experimental design with pre-test-post-test control group was used in the study. There were 26 participants in the experimental group and 29 participants in the control group of this study conducted in Amasya/Türkiye. In the study, high‑school students’ attitude scale towards programming and self-efficacy scale for programming for secondary‑school students were used. “Python Success Test” was prepared to measure students’ academic success in Python coding course. Separate ANCOVAs controlling for the corresponding pre-test scores showed no statistically significant differences between the AI-supported and traditional instruction groups in programming self-efficacy, attitude towards programming or academic achievement. The findings, therefore, do not support the superiority of either instructional approach. Further research in different settings and with larger samples is needed to identify the conditions under which teacher-guided chatbot use may support learning.