AI領航營建數位轉型：以ChatGPT輔助Excel VBA 工地照片管理系統開發實務
講者/Speaker： 林威延, Will Y. Lin
本研究探討如何利用大型語言模型（LLM）解決非資訊背景學生學習程式開發的痛點 。營建現地工程師常面臨極其繁瑣的表報管理工作，VBA 雖為自製自動化程式的一大利器，但傳統教學因語法艱澀、邏輯複雜，常使土木系學生產生挫折感 。本課程導入 LLM Coding 工具，使用「截圖輔助指令法」與「漸進式對話修正」教學。課堂以「工地照片管理系統」為範例，教學示範如何將原始碼截圖與具體需求組合成 AI 指令（Prompt），引導 AI 處理程式碼生成 ；同時刻意展示 AI 的錯誤（如直式照片縮放不當），示範如何透過迭代對話進行修正。結果顯示，在約 40 位修課學生中，高達 80% 的學生能獨立透過 AI 輔助，成功開發出具備自動置中與縮放功能的 VBA 程式。此模式有效降低了編程門檻，顯著提升學生的學習信心，使其能將專注力由「死背語法」轉移回「應用技術解決工程問題」的初衷，重塑數位工具開發之自信。
This study explores how Large Language Models (LLMs) can alleviate programming learning pains for non-computer science students. While VBA is a powerful tool for civil engineers to automate tedious construction site paperwork, traditional teaching methods often lead to frustration due to complex syntax and coding logic. This course innovatively implements LLM Coding tools through two newly designed methods: "Visual Prompting" (utilizing source code screenshots alongside textual instructions) and "Iterative Correction". Using a "Construction Site Photo Management System" as a case study, the instructor demonstrates how to combine code screenshots with functional requirements into precise AI prompts. Furthermore, the instructor intentionally displays AI imperfections (e.g., improper scaling of vertical photos) to teach students how to refine code through ongoing dialogue. The results show that 80% of the approximately 40 enrolled students successfully and independently developed a functional VBA program with automatic photo alignment and scaling features. This approach effectively lowers the entry barrier, boosts learning confidence, and allows students to focus on solving engineering problems rather than memorizing syntax, thereby reshaping their digital development efficacy.