paperbot · PL 论文追踪

全部论文 待读 收藏

P2 A Comparative Study on ChatGPT and Checklist as Support Tools for Unit Testing Education

ChatGPT 与检查清单作为单元测试教育支持工具的比较研究
FSE Proceedings of the 33rd ACM International Conference on the Foundations of Software Engineering2025匹配度 1/10 · 该论文属于软件工程教育研究,比较检查清单与 ChatGPT 对单元测试教学的辅助效果,不涉及静态分析、形式化验证、浮点程序分析、贝叶斯程序分析或 Agentic 程序分析等核心方向。其将 LLM 仅作为教学支持工具,而非用于改进程序分析或作为被分析对象,与用户研究画像基本无关。引用 4 inbox
Zihan Fang, Jiliang Li, Anda Liang, Gina R. Bai, Yu Huang
crossref: 2026-09-12

AI 速览

一句话:比较检查清单、ChatGPT 及二者结合对 42 名学生单元测试学习的影响,并给出教学建议。

单元测试教育中,已有轻量检查清单能改善学习结果,却难以帮助学生写出符合意图与设计的测试代码;生成式 AI(如 ChatGPT)被视为新的辅助工具。本文以 42 名学生为对象,比较检查清单、ChatGPT 及二者结合对单元测试表现的影响。结果显示,单独或组合使用这些工具对学生表现的影响相当;学生更偏好检查清单,但认可 ChatGPT 能加速任务完成并帮助解决编程语言问题。ChatGPT 虽具潜力,却未解决此前发现的实现挑战;对其依赖还可能使学生浅层处理 AI 生成内容、缺乏批判性评估,阻碍深入理解新概念。作者据此为学生和教师提出适应 AI 时代学习与教学的建议。

问题
已有检查清单能提升单元测试学习效果,但无法解决学生编写符合意图和设计的测试代码时的实现困难;同时 ChatGPT 等生成式 AI 工具作为新型支持工具的效果尚不明确。
方法
以 42 名学生为对象,设置检查清单、ChatGPT、二者结合等支持工具条件,比较其对单元测试表现和体验的影响。
评估
通过学生单元测试表现与反馈进行评估,发现各工具单独或组合使用效果相当;学生偏好检查清单,也认可 ChatGPT 的加速与语言帮助,但 ChatGPT 未克服实现挑战,并可能带来浅层学习与批判性思维不足。
原文摘要(Abstract)

Testing is widely practiced in software engineering, and many tools have been developed to support students in learning testing. Prior research suggests that a lightweight testing checklist improves learning outcomes but doesn't address students' challenges in writing test code that matches their intentions or design. Meanwhile, generative AI tools (e.g., ChatGPT) bring new promise as another form of software assistance tool. In this study, we examined the impact of various support tools (checklist, ChatGPT, or both) on unit testing among 42 students. Our results indicated that using these tools individually or in combination produced a comparable effect on student performance in unit testing. Students preferred using the checklist but acknowledged ChatGPT's effectiveness in accelerating task completion and addressing programming language challenges. While ChatGPT demonstrated potential benefits for testing education, it did not overcome the implementation challenges identified in the previous study. Moreover, reliance on ChatGPT may hinder students' deeper engagement with new concepts, which is crucial for comprehensive learning, as they often interact superficially with AI-generated responses without employing the critical thinking necessary to evaluate the information provided. Therefore, we proposed recommendations for both students and instructors on adapting to learning and teaching in the AI era and offer insights into the evolving role of AI in education.

链接与引用

DOI 原文 · PDF(开放获取) ·

BibTeX
@inproceedings{paperbot9083,
  title = {A Comparative Study on ChatGPT and Checklist as Support Tools for Unit Testing Education},
  author = {Zihan Fang and Jiliang Li and Anda Liang and Gina R. Bai and Yu Huang},
  booktitle = {FSE},
  year = {2025},
  doi = {10.1145/3696630.3727244}
}