學習問題拆解思維能力以提升人類與生成式人工智慧協作開發的程式碼品質;Learning Problem Decomposition Skills to Enhance Code Quality in Human-AI Collaborative Coding

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Computer Science and Information Engineering > Electronic Thesis & Dissertation > Item 987654321/95511

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/95511

Title:	學習問題拆解思維能力以提升人類與生成式人工智慧協作開發的程式碼品質;Learning Problem Decomposition Skills to Enhance Code Quality in Human-AI Collaborative Coding
Authors:	張文耀;Chang, Wen-Yao
Contributors:	資訊工程學系
Keywords:	生成式人工智慧;人機協作;運算思維;問題拆解;提示工程;電腦科學教育;Generative Artificial Intelligence;Human-AI Collaboration;Computational Thinking;Problem Decomposition;Prompt Engineering;Computer Science Education
Date:	2024-07-18
Issue Date:	2024-10-09 16:54:51 (UTC+8)
Publisher:	國立中央大學
Abstract:	隨著生成式人工智慧（Generative Artificial Intelligence, GAI）技術的飛速發展，軟體開發模式正經歷一場深刻的變革。從傳統的個人開發模式逐步轉向人機協作開發（Human-AI Collaborative Coding）模式，這一轉變迫使開發者必須掌握新的能力。許多研究指出，生成式 AI 在處理簡單且重複性高的任務（例如程式碼自動生成）上展現卓越的性能。因此，在人機協作開發中，為了最大化人工智慧的效能，開發者應學會將複雜問題盡可能拆解為更小、更明確定義的模組，使得人工智慧更容易生成正確的答案。接著分解其提示任務，以促進人工智慧達到最佳性能。這凸顯了問題拆解能力的重要性。本研究聚焦於探討開發者在人機協作開發中，Decomposition思維能力的重要性，並提出提示模板學習方法。透過引導學習者進行程式模組和任務拆解的訓練，旨在培養學習者的Decomposition思維能力，以提升人機協作開發的效果。本研究設計了一個學習系統，並進行了一項包含35名參與者、持續七週的學習實驗。研究結果顯示，學習者在採用提示模板學習方法後，在程式碼的模組化與可讀性方面達到了最佳的效果。參與者的反饋同樣表明，此學習方法能顯著提升程式碼品質。;With the rapid advancement of Generative Artificial Intelligence (GAI) technology, software development paradigms are undergoing profound transformations. Moving from traditional individual coding practices to Human-AI Collaborative Coding, this shift necessitates new capabilities among developers. Numerous studies have highlighted the exceptional performance of generative AI in handling simple and repetitive tasks, such as automatic code generation. Thus, in Human-AI collaborative coding, to maximize the efficiency of artificial intelligence, developers should learn to decompose complex problems into smaller, more clearly defined modules, enabling AI to generate more accurate solutions. Furthermore, refining prompt tasks is essential to optimize AI performance, highlighting the importance of decomposition skills. This study focuses on the significance of Decomposition thinking abilities in Human-AI collaborative coding and introduces a prompt template learning approach. By guiding learners through the training of program modularization and task decomposition, this approach aims to cultivate Decomposition thinking skills, thereby enhancing the effectiveness of Human-AI collaborative development. The study designed a learning system and conducted a seven-week experiment involving 35 participants. Results indicate that learners achieved optimal outcomes in code modularization and readability after adopting the prompt template learning method. Feedback from participants also demonstrates that this method significantly improves code quality.
Appears in Collections:	[Graduate Institute of Computer Science and Information Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	27	View/Open

社群 sharing

Loading...