LaBu：逢甲大學校園 AI 助教與 Vibe-Coding 學習平台
講者/Speaker： 蔡鈺鼎, Yu-Ting Tsai
本計畫以「一人一 AI 助教夥伴」為核心願景，致力於實踐教育資源平等的理念。儘管人工智慧技術日新月異，已成為現代學習不可或缺的關鍵工具，但校園場域中仍存在資源分配不均、平台功能分散以及使用門檻過高等現實挑戰。這些問題對於經濟弱勢學生的影響尤為顯著，高昂的商業 AI 服務費用往往成為他們獲取知識的重大阻礙。基於此背景，本計畫設計並開發了 LaBu——一個專為校園場景打造的 AI 助教平台。該平台打破了科系與經濟條件的限制，旨在為每一位學生提供穩定、可持續的 AI 助教體驗，進而有效縮小城鄉間與不同科系間的資源落差。
LaBu 的核心創新在於將 AI 從「被動的資訊提供者」深度轉化為「校園教學生態的關鍵成員」。系統的創新優勢具體體現在四個面向：第一，建立共享平台，不僅降低學生的學習負擔，更有效整合了校內的學習資源；第二，採用任務導向的工作流程設計，將複雜的大型任務拆解為可操作的子任務，引導學生在每個階段獲得恰當的 AI 指引。
在技術實作方面，系統以 JS 前端＋Python 後端（Flask Framework）為核心，串接大語言模型 API，並運用檢索增強生成（RAG）技術結合向量資料庫，實現專業領域知識的高效檢索與精準回應。平台採雲端部署模式，確保了系統的高可用性与可及性。本計畫的理論基礎融合格式化的 CDIO 教育模式與專案導向學習（Project-Based Learning），並注入 Vibe-Coding 的創新精神，提供了一套完善的 AI 輔助教學方案。研究結果證實，透過專屬校園場景的 AI 平台設計，不僅能顯著提升教學效率與學習成效，更能具體實踐教育資源平等的重大社會使命。
This project advances the vision of "one student, one AI teaching assistant companion," striving to promote educational resource equity. While artificial intelligence has become an essential tool for modern learning due to its rapid technological advancement, campus environments still encounter significant challenges, including uneven resource distribution, fragmented AI platforms, and high accessibility barriers. These issues disproportionately affect economically disadvantaged students, for whom the high costs of commercial AI service subscriptions serve as a major obstacle to access. To address these challenges, this project designed and developed LaBu, a campus-specific AI assistant platform. LaBu aims to break the barriers of academic majors and financial circumstances, providing every student with a stable, sustainable AI companion, and thereby helping to bridge the digital divide between urban and rural areas, as well as across different disciplines.
The core innovation of this research lies in transforming AI from a "passive information provider" into an integral component of the campus teaching ecosystem. The system features four key innovations: First, it establishes a university-wide shared platform that alleviates economic burdens while consolidating campus learning resources. Second, it adopts a task-oriented workflow design that breaks down complex objectives into manageable sub-tasks, guiding students through a step-by-step process with contextually appropriate AI assistance.
From a technical perspective, the system is built upon a JS frontend + Python backend (Flask Framework), integrates Large Language Model APIs, and employs Retrieval-Augmented Generation (RAG) combined with a vector database for efficient professional knowledge retrieval and precise response generation. The platform utilizes a cloud-based deployment model to ensure high stability and accessibility. Grounded in the CDIO educational methodology and Project-Based Learning principles, and infused with the spirit of Vibe-Coding, this project offers a comprehensive AI-assisted teaching framework. The research demonstrates that a purpose-built campus AI platform can not only effectively enhance teaching efficiency and learning outcomes but also fulfill the critical goal of educational resource equity.
- 論文全文：LaBu：逢甲大學校園 AI 助教與 Vibe-Coding 學習平台
- 影像： 相關影片