摘要
生成式人工智能(Generative Artificial Intelligence,GenAI)技术以其独特的生成性、创造性和适应性对教学和学习产生了重大影响,但其对学生学习效果的实际作用存在争议,制约了其在教育实践的广泛应用。文章采用元分析方法,对2024年6月前有关GenAI影响学生学习效果的39篇实验和准实验研究文献(60个效应量)进行了系统梳理与分析,结果表明:GenAI对学生学习效果具有中等偏大的积极影响(g=0.623);与其他学科相比,GenAI更能提升医学学科的学习效果;GenAI对学生学习效果的影响随着样本量增加而减弱;持续多次使用GenAI比一次性干预更能促进学生学习,但不同干预时长下无显著差异;相较于内容生成、评估反馈和混合应用方式,会话交互式GenAI应用最广且最有效;GenAI适用于各学段和不同学习者群体。基于上述结果,文章建议从学科特色、场景差异、合理干预、情境适配等方面加强对学生学习效果的实际作用,以助力GenAI教育应用落地。
Generative artificial intelligence(GenAI)technology has demonstrated a significant impact on teaching and learning due to its unique generativity,creativity,and adaptive properties.However,its practical effect on students’learning outcomes is controversial,thrreby limiting its widespread application in educational practice.A meta-analysis method was used to systematically sort out and analyze 39 experimental and quasi-experimental articles(60 effect sizes)on the influence of GenAI on students’learning outcomes published prior to June 2024.The analysis revealed several key findings:GenAI exhibited a moderately positive effect on students’learning outcomes(g=0.623);GenAI can better enhance the learning outcomes of medical disciplines compared to other disciplines;The influence of GenAI on students’learning outcomes weakened with the sample size increase;Continuous and multiple uses of GenAI was more effective in enhancing students’learning than one-time intervention,but there was no significant difference across different intervention durations;Conversational interaction modalities outperformed other application types(content generation,assessment feedback,and blended approaches)in both prevalence and effectiveness;GenAI was sutiable for all educational stages and different learner groups.Based on the above results,this study suggested to strengthen the practical effect on students’learning outcomes from the aspects of subject characteristics,scene differences,reasonable intervention,and context adaptation to facilitate the implementation of GenAI educational applications.
作者
钱莉
李文昊
顾庭轩
王小寒
QIAN Li;LI Wen-Hao;GU Ting-Xuan;WANG Xiao-Han(Faculty of Artificial Intelligence in Education,Central China Normal University,Wuhan,Hubei,China 430079)
出处
《现代教育技术》
2025年第8期36-45,共10页
Modern Educational Technology
基金
湖北省高等学校省级教学研究项目“数据驱动的异步在线讨论行为特征分析与学习策略研究”(项目编号:2023089)
华中师范大学优秀研究生教育创新资助项目“VR环境下生成式AI驱动的师范生提问能力提升研究”(项目编号:2025CXZZ035)的阶段性研究成果。
关键词
生成式人工智能
学习效果
元分析
实验研究
generative artificial intelligence
learning effect
meta-analysis
experimental study
