Abstract
Python programming students often hit errors outside office hours, and traditional TA capacity cannot cover demand. We build a generative-AI tutoring system specialized for instrument-automation programming, providing real-time debugging guidance and conceptual scaffolding. Pre/post evaluation in a course on instrument automation shows measurable skill gains.
Problem & Motivation
Python programming students frequently hit errors outside office hours with no one to ask. Traditional TA capacity does not scale, and students working on domain-specialized applications such as instrument automation need help that goes beyond general programming.
Method
We developed a generative-AI tutoring system specialized for instrument-automation programming. The system responds to student questions in real time, offers debugging suggestions, and walks students through underlying program logic. We evaluated impact through pre/post tests and surveys.
Findings
- Measurable gains in instrument-automation programming skill.
- Students reported stronger autonomous learning behavior and reduced dependence on human TAs.
- Strong positive ratings for debugging assistance and conceptual explanations.
Implications
AI tutors can complement — not replace — human teaching support, particularly outside synchronous hours. For applied programming courses spanning multiple disciplines, the ability to customize the tutor's domain knowledge is decisive: a general tutor will not match a tutor tuned to the actual application context students face.
Citation
BibTeX
@inproceedings{phay2025ai_assistant_python,
author = {Kar Sue Phay and Chia-Kai Chang},
title = {A Generative Artificial Intelligence-Based {Python} Learning Assistance System: Assessing Its Impact on Instrument Automation Skills},
booktitle = {Proc. IEEE Conf. on Education and Training in Optics and Photonics (ETOP)},
pages = {1--4},
year = {2025},
month = may,
doi = {10.1109/ETOP64842.2025.11030677},
}