夜雨聆风AI进课堂后，真正难的不是工具，而是落地 480篇研究总结：生成式AI教育应用的机会、风险与回应 Based on: García-López, Ramírez-Montoya, & Molina-Espinosa (2026), Interactive Learning Environments |
核心一句话 这篇系统综述的重点不是“AI到底好不好”，而是：生成式AI带来个性化学习与教育创新的机会，但真正落地需要政策、学校制度、伦理监管和法律适配一起跟上。 |
Generative AI is already changing education.生成式AI已经在改变教育。
Students use ChatGPT, DeepSeek, Doubao, and other tools to summarize readings, draft essays, revise presentations, generate ideas, and prepare for exams.学生用 ChatGPT、DeepSeek、豆包等工具总结阅读、起草论文、修改展示、生成思路、准备考试。
Teachers use them to design activities, generate examples, create feedback, and explore new teaching formats.教师用它们设计活动、生成案例、制作反馈、探索新的教学形式。
Schools and universities now face a harder question: how should GenAI be integrated into teaching, assessment, institutional policy, and educational regulation?学校和大学现在面对的是一个更难的问题：生成式AI应该如何进入教学、评价、学校政策和教育监管？
A systematic review published in Interactive Learning Environments gives a useful answer.一篇发表于 Interactive Learning Environments 的系统综述给出了一个有用答案。
The authors reviewed GenAI education research from 2019 to 2023. They searched Scopus and Web of Science, retrieved 801 articles, and finally included 480 studies.作者系统分析了2019年至2023年间的生成式AI教育研究。他们从 Scopus 和 Web of Science 检索到801篇文章，最终纳入480篇研究。
The paper identifies three major constructs: personalization of learning, educational innovation and motivation, and ethical and inclusive regulation.论文归纳出三个核心构念：个性化学习、教育创新与学习动机，以及伦理与包容性监管。
Key takeaway / 关键判断 GenAI offers real opportunities for personalized learning and educational innovation, but responsible implementation depends on policies, institutional guidelines, and legal adaptation. 生成式AI确实能为个性化学习和教育创新提供机会，但负责任的落地依赖政策、学校制度和法律适配。 |
1. 这篇综述怎么做？ 1. How was this review conducted? |
This is a systematic literature review, not a single classroom experiment.这是一篇系统文献综述，不是单个课堂实验。
The authors mainly did five things:作者主要做了五件事：
Step / 步骤 | What the paper did / 论文做了什么 |
Search / 检索 | Scopus and Web of Science, 2019-2023 Scopus 和 Web of Science，2019-2023 |
Screen / 筛选 | 801 records retrieved; duplicates and irrelevant studies removed 初始获得801篇，去除重复和无关研究 |
Include / 纳入 | 480 studies entered the final review 最终480篇进入系统综述 |
Code / 编码 | Studies were classified by research questions and themes 按研究问题和主题对文献分类 |
Analyze / 分析 | Excel, Rayyan, NVivo 14, Python, thematic mapping, and visualizations 使用 Excel、Rayyan、NVivo 14、Python、主题映射和可视化 |
The coding process was checked for reliability. Two independent researchers coded a random 20% sample, and Cohen’s Kappa reached 0.87.编码过程也进行了可靠性检验。两位独立研究者对随机抽取的20%样本文献进行编码，Cohen’s Kappa 达到0.87。
Evidence note / 证据说明 它不能证明每一种生成式AI工具都能提升学习。它展示的是：这个领域正在关注什么，哪些机会被反复讨论，哪些风险仍未解决。 |
2. 当前研究最关注什么？ 2. What does current GenAI education research focus on? |
The paper reports four main thematic areas.论文报告了四个主要主题领域。
Theme / 主题 | Share / 占比 | Main focus / 主要内容 |
Personalization of learning 个性化学习 | 40% | Adaptive learning, real-time feedback, individualized learning paths 自适应学习、实时反馈、个别化学习路径 |
Educational innovation 教育创新 | 35% | Gamification, simulations, immersive environments, interactive learning 游戏化、模拟学习、沉浸式环境、互动学习 |
Ethical regulation 伦理监管 | 15% | Bias, data security, transparency, fairness 偏见、数据安全、透明度、公平 |
Policy frameworks 政策框架 | 10% | Institutional policies, global standards, legal and organizational responses 学校政策、全球标准、法律与组织回应 |
Key takeaway / 关键判断 AI education research is currently application-heavy, but governance is still catching up. 当前AI教育研究更重应用，治理仍在追赶。 |
3. 机会一：个性化学习 3. Opportunity one: personalized learning |
Personalization is the largest theme in the review.个性化学习是这篇综述中占比最高的主题。
GenAI can adapt content, rhythm, examples, feedback, and support to different students.生成式AI可以根据不同学生调整内容、节奏、例子、反馈和支持方式。
For teachers, this matters because large classes make individualized support difficult.对教师来说，这一点很有价值，因为大班教学很难持续提供个别化支持。
Function / 功能 | Educational use / 教育用途 |
Adaptive explanation 自适应解释 | Explain the same concept at different difficulty levels 用不同难度解释同一个概念 |
Real-time feedback 实时反馈 | Give immediate comments on drafts, answers, or exercises 对草稿、答案或练习即时反馈 |
Diagnostic support 诊断支持 | Identify weak areas before students fall behind 在学生落后前识别薄弱点 |
Personalized path 个性化路径 | Recommend materials or tasks based on student needs 根据学生需求推荐材料或任务 |
Formative assessment 形成性评价 | Support low-stakes checking during learning 在学习过程中进行低风险检测 |
But personalization also creates risks. The paper repeatedly connects personalization with data privacy, algorithmic bias, transparency, and equity.但个性化学习也带来风险。论文反复把个性化学习与数据隐私、算法偏见、透明度和公平联系起来。
Practical question / 实操问题 Before using AI for personalized learning, do we know what student data are collected, how recommendations are generated, and who can inspect or challenge the results? 在使用AI进行个性化学习前，我们是否知道系统收集了哪些学生数据、推荐如何生成、谁可以检查或质疑结果？ |
4. 机会二：教育创新与学习动机 4. Opportunity two: educational innovation and motivation |
The second largest theme is educational innovation.第二大主题是教育创新。
The paper discusses gamified platforms, simulations, virtual environments, intelligent tutoring systems, and interactive learning spaces.论文讨论了游戏化平台、模拟学习、虚拟环境、智能辅导系统和互动学习空间。
These tools may increase motivation and engagement.这些工具可能提升学习动机和参与度。
GenAI application / 生成式AI应用 | What it changes / 它改变了什么 |
AI simulations AI模拟学习 | Students explore complex scenarios interactively 学生可以互动探索复杂情境 |
Gamified learning 游戏化学习 | Tasks become adaptive, staged, and feedback-rich 任务变得自适应、有阶段、有反馈 |
Virtual assistants 虚拟助手 | Students receive on-demand support 学生可以获得即时支持 |
AI-supported projects AI支持的项目学习 | Students brainstorm, test ideas, and revise outputs 学生可以头脑风暴、测试想法、修改产出 |
Immersive environments 沉浸式环境 | Abstract concepts become more concrete and explorable 抽象概念变得更具体、可探索 |
The paper also gives a caution: innovation must remain aligned with pedagogical objectives.论文也提出提醒：创新必须始终服务教学目标。
Teacher test / 教师可以用一个简单测试 After using AI, are students only producing more outputs, or are they explaining, evaluating, applying, and reflecting better? 使用AI之后，学生只是生成了更多产出，还是能更好地解释、评价、应用和反思？ |
5. 挑战一：伦理与包容性监管 5. Challenge one: ethical and inclusive regulation |
The paper treats ethical regulation as a central condition for responsible GenAI use.论文把伦理监管视为负责任使用生成式AI的核心条件。
Ethical issue / 伦理问题 | Why it matters / 为什么重要 |
Transparency 透明度 | Students and teachers should know how AI is used 学生和教师应知道AI如何被使用 |
Fairness 公平性 | AI should not reinforce unequal access or biased outcomes AI不应强化不平等接入或有偏结果 |
Data protection 数据保护 | Student data are sensitive and require safeguards 学生数据敏感，需要保护机制 |
Algorithmic bias 算法偏见 | AI outputs may reflect biased data or design assumptions AI输出可能反映有偏数据或设计假设 |
Inclusion 包容性 | Policies must consider different regions, resources, and learners 政策需考虑不同地区、资源和学习者 |
Education is not only about efficiency. It also involves rights, opportunities, assessment, and fairness.教育不只是效率问题，它还涉及权利、机会、评价和公平。
Institutional question / 机构可以追问 Does our AI policy protect students who have less access, lower AI literacy, or weaker technological resources? 我们的AI政策是否保护了那些技术接入不足、AI素养较低或资源较弱的学生？ |
6. 挑战二：学校制度政策 6. Challenge two: institutional policies |
The paper identifies institutional policy as a major gap.论文指出，学校制度政策是一个重要缺口。
Institutional area / 学校政策领域 | What schools need to clarify / 学校需要说清楚什么 |
Teacher training 教师培训 | How teachers learn to use GenAI responsibly 教师如何学习负责任使用生成式AI |
Ethical guidelines 伦理指南 | What counts as acceptable and unacceptable use 什么是可接受和不可接受的AI使用 |
Impact assessment 影响评估 | How institutions evaluate AI’s effects on learning 学校如何评价AI对学习的影响 |
Academic data protection 学术数据保护 | How student data are stored, used, and protected 学生数据如何存储、使用和保护 |
Assessment rules 评价规则 | How AI use should be disclosed and evaluated AI使用如何披露、如何评价 |
GenAI tools are already being used, but many institutions have not yet built clear rules.生成式AI工具已经在被使用，但很多机构还没有建立清晰规则。
Question / 问题 | Why it matters / 为什么重要 |
When can students use GenAI? 学生什么时候可以使用生成式AI？ | Reduces ambiguity 减少模糊地带 |
When must students disclose AI use? 学生什么时候必须披露AI使用？ | Supports academic integrity 支持学术诚信 |
What data can AI tools collect? AI工具可以收集哪些数据？ | Protects privacy 保护隐私 |
Who evaluates AI tools before adoption? 谁在采用前评估AI工具？ | Reduces institutional risk 降低学校风险 |
What training do teachers receive? 教师接受什么培训？ | Supports effective implementation 支持有效落地 |
7. 挑战三：法律监管 7. Challenge three: legal regulation |
The paper’s conclusion is cautious: current legal frameworks may not fully address the challenges posed by GenAI in education.论文的结论较为谨慎：现有法律框架可能还无法充分应对生成式AI在教育中带来的挑战。
Legal concern / 法律关注 | Education-related question / 教育中的问题 |
Data privacy 数据隐私 | Can student data be used by AI platforms? 学生数据能否被AI平台使用？ |
Academic integrity 学术诚信 | How should AI-assisted work be regulated? AI辅助作品应如何监管？ |
Accountability 责任归属 | Who is responsible for harmful or biased AI output? 有害或有偏AI输出由谁负责？ |
Intellectual rights 知识产权 | Who owns AI-assisted educational content? AI辅助生成的教学内容归谁所有？ |
International coordination 国际协作 | How can rules work across platforms and regions? 不同平台和地区之间规则如何协调？ |
GenAI changes quickly. Static rules may become outdated quickly. Education-specific AI regulation therefore needs regular review.生成式AI变化很快，静态规则容易过时。因此，面向教育的AI监管需要定期更新。
Policy question / 政策问题 Which current rules were designed for older forms of academic misconduct, data use, or digital tools — and no longer fit GenAI? 哪些现有规则是为旧形式的学术不端、数据使用或数字工具设计的，已经不再适配生成式AI？ |
8. 学校和教师可以带走什么？ 8. What should schools and teachers take away? |
The paper’s five research directions can be turned into five practical questions.论文的五个研究方向，可以转化为五个实践问题。
Paper direction / 论文方向 | Practical question / 实践问题 |
GenAI competencies 生成式AI能力 | What AI skills do students and teachers need? 学生和教师需要哪些AI能力？ |
Creative use 创造性使用 | How can AI support innovation without replacing learning? AI如何支持创新，而不是替代学习？ |
Policy responses 政策回应 | What rules are needed for academic integrity and responsible use? 学术诚信与负责任使用需要哪些规则？ |
Institutional policies 学校制度政策 | How will schools support teacher training and data protection? 学校如何支持教师培训和数据保护？ |
Legal regulation 法律监管 | How should existing laws adapt to GenAI risks? 现有法律如何适配生成式AI风险？ |
For teachers, the immediate task is not only learning better prompts. It is learning how to design AI-supported activities with clear learning goals.对教师来说，眼前任务不只是学习更好的提示词，更重要的是学会如何围绕明确学习目标设计AI支持的活动。
For schools, the task is not only choosing a platform. It is building rules for disclosure, assessment, data protection, teacher training, and equity.对学校来说，任务不只是选择某个平台，更重要的是建立关于使用披露、评价、数据保护、教师培训和公平性的规则。
For students, the task is not only knowing how to use AI. It is knowing when AI is helping them learn and when it is replacing their own work.对学生来说，任务不只是知道如何使用AI，更重要的是知道什么时候AI是在帮助自己学习，什么时候AI正在替代自己的工作。
9. 所以呢？ 9. So what? |
The paper gives a balanced view of GenAI in education.这篇论文提供了一种平衡视角。
It does not say GenAI should simply be banned. It also does not say GenAI should be adopted without concern.它没有说生成式AI应该被简单禁止，也没有说生成式AI可以毫无顾虑地采用。
Its message is: GenAI can support personalized learning, innovation, motivation, and teaching efficiency, but these benefits depend on responsible implementation.它的信息是：生成式AI可以支持个性化学习、教育创新、学习动机和教学效率，但这些收益取决于负责任的落地。
Requirement / 要求 | Meaning / 含义 |
AI competencies AI能力 | Students and teachers need technical, critical, and pedagogical skills 学生和教师需要技术性、批判性和教学性能力 |
Pedagogical design 教学设计 | AI use must serve learning goals AI使用必须服务学习目标 |
Ethical regulation 伦理监管 | Transparency, fairness, privacy, and bias must be addressed 必须处理透明度、公平、隐私和偏见 |
Institutional policy 学校政策 | Schools need clear rules and teacher training 学校需要清晰规则和教师培训 |
Legal adaptation 法律适配 | Existing regulations need updates for GenAI challenges 现有监管需要回应生成式AI挑战 |
Key takeaway / 关键判断 The better question is not: Is GenAI good or bad for education? The better question is: What conditions are needed for GenAI to be used responsibly in education? 更好的问题不是：生成式AI对教育是好是坏？而是：教育中负责任使用生成式AI，需要哪些条件？ |
Discussion question / 讨论问题 Among the five dimensions in this paper — AI competencies, creative use, policy responses, institutional policies, and legal regulation — which one is currently the weakest in your school or institution? 在这篇论文讨论的五个维度中——AI能力、创造性使用、政策回应、学校制度政策和法律监管——你认为你所在学校或机构目前最薄弱的是哪一个？ 欢迎在评论区聊聊。 |
10. 参考文献 10. Reference |
García-López, I. M., Ramírez-Montoya, M. S., & Molina-Espinosa, J. M. (2026). Generative artificial intelligence in education: A systematic analysis of opportunities, challenges, and responses. Interactive Learning Environments, 34(3), 1194–1217. https://doi.org/10.1080/10494820.2025.2519133
#AI教育 #生成式AI #教育技术 #高等教育 #ChatGPT #DeepSeek #豆包 #个性化学习 #教育创新 #AI伦理 #教育治理 #学术诚信
