摘要
个性化作业是落实“双减”政策,实现减负提质的重要途径。人工智能为实现个性化作业提供了技术支撑,但在实践中,由于其内部工作原理对用户不可见,且存在价值观、情感等方面的缺失问题,降低了其在个性化作业实践中应用的深度与效果。人在回路机制是指通过置信度评估的方法对人工智能算法生成的结果进行评估,进而决定是否执行算法结果的人机协同方法。基于人在回路机制的个性化作业设计理论模型,以作业实践为核心流程,以人机协同、认知诊断、干预反馈等为理论指导,将人在回路机制融入作业布置、作业诊断和作业辅导的关键环节中,实现了人类智能与人工智能的协同工作。在此基础上提出的个性化作业实践路径,包含基于学习者画像的作业习题推荐、基于多模态学习分析技术的错因诊断,以及基于多模态资源推送的作业辅导三个阶段,分别回答了做什么题、错在哪里,以及如何辅导等作业实践中的关键问题,形成了人机协同机制下智能技术赋能作业实践的闭环。基于该实践路径研发的原型系统,已在一线教学单位开展了实践,初步验证了方法的可行性。
Personalized homework is an important way to implement the“double reduction”policy to achieve burden reduction and quality improvement.Artificial intelligence provides technical support for realizing personalized homework,but in practice,its application in personalized homework is limited in depth and effectiveness due to its invisible internal working principles,and the lack of values and emotions.The human-in-the-loop mechanism refers to the method of evaluating the results generated by artificial intelligence algorithms through confidence assessment,and then determining whether to implement the algorithm results in a human-machine collaborative approach.The theoretical model of personalized homework design based on the human-in-the-loop mechanism takes homework practice as the core process,and takes human-machine collaboration,cognitive diagnosis,intervention feedback and so on as theoretical guidance.It integrates the human-in-the-loop mechanism into crucial links such as homework assignment,homework diagnosis,and homework tutoring,realizing the collaboration of human intelligence and artificial intelligence.Based on this,the practical path of personalized homework proposed in this study contains three stages:homework recommendation based on learner personas,error diagnosis based on multi-modal learning analysis technology,and homework tutoring based on multi-modal resources.These stages respectively respond to key questions in homework practice such as what problems to solve,where the errors are,and how to tutor,forming a closed loop of intelligent technology empowering homework practice under the human-machine collaborative mechanism.The prototype system developed based on this practical path has been carried out in the first-line teaching units,preliminarily verifying the feasibility of the method.
出处
《现代远程教育研究》
CSSCI
北大核心
2024年第1期93-100,共8页
Modern Distance Education Research
基金
国家社会科学基金2020年度教育学一般课题“基于人机智能协同的精准学习干预研究”(BCA200080)。
关键词
双减
个性化作业
作业设计
人在回路
人机协同
Double Reduction
Personalized Homework
Homework Design
Human-in-the-Loop
Human-Machine Collaboration
作者简介
武法提,教授,博士生导师,北京师范大学数字学习与教育公共服务教育部工程研究中心(北京100875);李坦,博士研究生,北京师范大学教育学部(北京100875);任伟祎,硕士研究生,北京师范大学教育学部(北京100875);夏志文,硕士研究生,北京师范大学教育学部(北京100875)。
