一句话：基于多利益相关者调查揭示 GenAI 在软件工程教育与实践中的错位，并提出三层生态系统对齐框架。
生成式 AI 正在快速重塑软件工程教育与专业实践，但其在学习、教学与产业中的采用并不均衡且缺乏协调。本文通过对学生、教师与从业者进行多利益相关者问卷调查，考察 GenAI 在不同教育与专业情境中的学习与使用方式。结果发现：各方普遍依赖自主学习，机构支持有限，且存在角色特定的伦理顾虑，从而揭示出教育与实践之间的结构性错位。为应对这一挑战，作者提出一个三层生态系统框架，将 GenAI 的使用与软件工程职业生命周期及跨角色交互对齐。
Generative AI (GenAI) is rapidly reshaping software engineering (SE) education and professional practice, yet its adoption across learning, teaching, and industry remains uneven and only partially aligned. Students widely use GenAI for programming tasks, teachers integrate it cautiously due to academic integrity concerns, and professional developers increasingly rely on it under organizational and governance constraints. This paper reports findings from a multi-stakeholder survey of students, teachers, and professionals, examining how GenAI is learned and used across educational and professional contexts. The results indicate widespread reliance on self-directed learning, limited institutional support, and role-specific ethical concerns, suggesting a structural misalignment between education and practice. To address this challenge, we propose a three-layer ecosystem framework that aligns GenAI use in SE with professional lifecycles and cross-role interactions.