人智协同问题解决能力测评:基于ECD方法实现
文 桥1, 2 马志强1, 2 乔子涵1, 2
(1.江南大学 教育学院,江苏,无锡 214122;2.江南大学 江苏“互联网+教育”研究基地,江苏 无锡 214122)
【摘 要】随着生成式人工智能技术的演化迭代与场景化落地应用,人与智能体协同解决问题能力的重要性日益凸显。如何对上述能力进行精准评测成为教育领域亟待解决的重要问题。研究基于证据为中心的设计方法,首先,构建了由人智协同与问题解决两个过程交叉形成的能力模型;其次,设计了包含劣构问题情境、结构化任务序列等特征的任务模型;最后,采用“关键节点取证、等级量规定分、多源证据加权”的操作路径建立证据模型,由此形成人智协同问题解决能力的测评框架。针对大学生群体的初步应用与可行性检验表明,该框架具有较好的评分者一致性和结果解释价值,可为个体能力剖面生成与群体能力诊断提供依据。未来研究需关注能力模型的动态演化迭代、任务模型的生成式适配以及证据模型的人机协同评分机制,从而推动测评从实验情境走向真实教育场景。
【关键词】人智协同;人智协同问题解决能力;问题解决能力;证据中心设计;协作问题解决
Assessing Human-Agent Collaborative Problem-Solving Competence: An Evidence-Centered Design Approach
WEN Qiao1, 2, MA Zhiqiang1, 2 and QIAO Zihan1, 2
(1. College of Education, Jiangnan University, Wuxi 214122, China; 2. Jiangsu “Internet plus Education” Research Center, Jiangnan University,Wuxi 214122, China)
Abstract: With the iterative evolution and scenario-based application of generative artificial intelligence technologies, the importance of the ability to solve problems collaboratively with intelligent agents has become increasingly prominent. How to accurately assess this ability has become an urgent issue that needs to be addressed in the field of education. Grounded in the evidence-centered design (ECD) approach, this study first constructs a competence model formed by the intersection of two processes—human–agent collaboration and problem solving; second, it designs a task model featuring ill-structured problem scenarios and structured task sequences; finally, it establishes an evidence model through an operational pathway of “collecting evidence at key nodes, scoring with rating scales, and weighting multi-source evidence”, thereby forming an assessment framework for human–agent collaborative problem-solving competence. Preliminary application and feasibility testing among university students indicate that the framework demonstrates satisfactory inter-rater consistency and value in interpreting results, and provide a basis for generating individual competence profiles and diagnosing group competence. Future research could attend to the dynamic evolution and iteration of the competence model, the generative adaptation of the task model, and the human–machine collaborative scoring mechanism of the evidence model, so as to advance assessment from experimental settings to real educational contexts.
Keywords: human-agent collaboration; human-agent collaborative problem-solving competence; problem-solving competence; evidence-centered design; collaborative problem solving
下载:
人智协同问题解决能力测评:基于ECD方法实现.pdf