【摘要】 以ChatGPT、DeepSeek为代表的生成式人工智能在教育领域广泛应用,但其对学生自我调节学习能力的影响效果仍存疑,缺乏系统性综述论证。基于此,采用元分析的方法,通过对20项实验和准实验研究进行分析,发现生成式人工智能整体上对学生自我调节学习能力有着中等程度的正向影响作用,且在目标设定、学习策略、时间管理、寻求帮助和自我评价五大关键维度均有促进作用。调节变量显示,不同教育阶段生成式人工智能干预效果呈非线性差异,小学和高等教育阶段的效应显著高于中学,短期干预的成效显著高于长期干预,对话生成类工具的支持效应最为显著。基于研究结果,建议未来要从控制实验干预时长、优化接入策略、强化交互以及结合学科特点来部署生成式人工智能在教育中的应用。更多还原

【Abstract】 With the widespread application of generative artificial intelligence(GenAI) in education—represented by tools such as ChatGPT and DeepSeek—its impact on students’ self-regulated learning(SRL) ability remains a topic of debate, lacking systematic synthesis and empirical validation. To address this gap, this study employed a meta-analytic approach to examine 20 experimental and quasi-experimental studies. The results reveal that GenAI exerts a moderately positive overall effect on students’ SRL and promotes improvement across five key dimensions: goal setting, learning strategies, time management, help-seeking, and self-evaluation. Moderator analyses indicate nonlinear differences in intervention effects across educational stages, with primary and higher education showing significantly stronger effects than secondary education. Moreover, short-term interventions yield greater effectiveness than long-term ones, and dialogic generation tools demonstrate the most substantial supportive impact. Based on these findings, it is recommended that future research and practice focus on controlling intervention duration, refining GenAI integration strategies, enhancing human-AI interaction, and aligning applications with subject-specific characteristics.更多还原