"PLANATHON ACTIVITY - STUDENT FEEDBACK SURVEY"

"The integration of generative AI tools like large language models (LLMs) into software engineering (SE) curricula presents a challenge: balancing efficiency gains with the preservation of critical thinking and manual learning processes. This study proposes a two-phase instructional model that begins with manual project planning tasks (e.g., SRS, WBS, estimation, scheduling) and transitions into AI-assisted application through a capstone-style simulation called \"Planathon.\" Implemented in an undergraduate-level software project management course (N = 57), the model guided students through foundational skill building before introducing LLMs (e.g., ChatGPT, Gemini) and task-specific AI tools. Findings show strong performance in estimation and WBS tasks but reveal common challenges in diagram generation and scheduling. Student reflections revealed effective prompt use and AI literacy, though limited engagement with ethical considerations. The results validate a scaffolded approach to AI integration\u2014preserving depth while enhancing productivity\u2014and offer a replicable framework for preparing SE students with critical, industry-aligned AI competencies."

生成式AI（Generative AI）工具（如大语言模型（Large Language Model，LLM））融入软件工程（Software Engineering，SE）课程体系面临一项核心挑战：需在提升教学效率与保留批判性思维培养、手动实操学习流程之间达成平衡。本研究提出一种两阶段教学模式：先开展手动项目规划任务（例如软件需求规格说明书（Software Requirements Specification，SRS）撰写、工作分解结构（Work Breakdown Structure，WBS）搭建、工作量估算、进度排程），再通过名为“Planathon”的顶石式模拟项目过渡至人工智能辅助应用环节。该模式在一门本科软件工程项目管理课程（样本量N=57）中实施，教学流程为先引导学生夯实基础技能，再引入大语言模型（例如ChatGPT、Gemini）及面向特定任务的人工智能工具。研究结果显示，学生在工作量估算与工作分解结构搭建任务中表现优异，但在图表生成与进度排程环节普遍存在难点。学生反思反馈显示，其已掌握高效的提示词（Prompt）使用技巧与人工智能素养，但在伦理考量方面的参与度相对有限。研究结果证实了脚手架式人工智能融入教学路径的有效性：既保留了学习深度，又提升了学习效率，并为培养符合行业需求、兼具批判性思维与人工智能胜任力的软件工程专业学生提供了可复制的实施框架。