摘要
教师开展循证实践面临诸多挑战,包括难以获取和利用证据、缺乏有效的反馈和评价机制,以及实践过程中指导和支持不足,大模型技术的发展为破解这些难题带来了机遇。本研究构建了大模型驱动的循证实践行动框架,涵盖循证实践的五个关键阶段:提出问题、检索证据、评价证据、应用证据和后效评估,并针对不同阶段的特点设计了相应的大模型支持策略;开发了GPT4EBP平台,为教师提供基于自然语言交互的个性化指导、反馈和评估,还针对师范生开展了为期六周的循证实践。问卷调查结果表明,借助GPT4EBP平台,师范生能更高效、高质量地完成循证实践任务,其循证实践能力和集体效能感均得到显著提升;访谈数据揭示了平台在简化流程、提供及时反馈等方面的价值。本研究揭示了大模型技术在优化循证实践流程、赋能教师专业决策等方面的独特价值,丰富和发展了循证实践的理论体系,相应的行动框架和GPT4EBP平台可为教师循证实践提供个性化支持,使其能便捷获取证据、系统分析证据并将证据转化为可落地的教学行动,提升循证实践的针对性和有效性。
Teachers face numerous challenges in evidence-based practices(EBP).This study explores how to integrate large language models into the EBP process for teachers by constructing an action framework.This framework encompasses the five key stages of EBP:Posing questions,retrieving evidence,evaluating evidence,applying evidence,and post-implementation assessment,and designs specific large language model support strategies for different stages.Furthermore,a platform named GPT4EBP was developed by utilizing a large model agent architecture to allow teachers to interact in natural language to offer personalized guidance,feedback,and evaluation for all stages of evidencebased practice.To test the effects of the framework and GPT4EBP platform,a case study was conducted with pre-service teachers.Over a six weeks,the 30 graduate students in teacher education programs used the GPT4EBP platform to carry out EBP to optimize existing lesson plans in groups.The results indicated that the students were able to complete their EBP tasks more efficiently and with higher quality,and significantly enhanced their skills in EBP and collective efficacy.Interview data further revealed the unique value of the platform in simplifying processes and in providing timely feedback.The study demonstrates the unique value of large language model technology in optimizing EBP processes and in empowering decision-making for teaching.It enriches and expands the knowledge of EBP,the action framework,and GPT4EBP platform that provide personalized support at each stage of the EBP with easy access to evidence,and systematic analysis of evidence,and transforming evidence into actionable teaching strategies to enhance evidence-based practice.
作者
褚乐阳
刘泽民
王浩
陈向东
CHU Leyang;LIU Zemin;WANG Hao;CHEN Xiangdong(Department of Educational Information Technology,East China Normal University,Shanghai 200062,China)
出处
《开放教育研究》
CSSCI
北大核心
2024年第4期91-103,共13页
Open Education Research
基金
全国教育科学规划一般课题“基于大语言模型的青少年人工智能教育研究”(BCA230276)
华东师范大学上海智能教育研究院2023年智能教育博士生科研基金“基于大模型的人机协同循证实践平台的开发与实践应用”(202301002)。
关键词
教师专业发展
循证实践
大模型
行动框架
GPT4EBP
teacher professional development
evidence-based practice
large language model
action framework
