建筑材料课程人机协同教学的认知卸载调控

Regulating Cognitive Offloading in Human-AI Collaborative Teaching of Building Materials Courses

摘要: 文章针对建筑材料课程人机协同教学中认知卸载既能释放认知资源，又可能削弱学生独立判断与迁移能力的双重效应，采用文献分析、逆向教学设计和“目标–任务–证据”一致性审查，以滨海低碳混凝土配合比设计为例，明确人工智能的使用权限与学生应承担的认知责任。在此基础上，提出“独立预判–有界推演–专业核验–支架撤除–无AI迁移”五阶段调控链，设置基线、过程和终点三类检验，并从独立迁移、依赖落差、核验质量、信心校准和延迟保持等方面构建评价指标。该方案并非排斥人工智能，而是选择性卸载信息检索、重复计算等次要活动，同时保留问题界定、机理解释、标准判定和工程决策等核心认知责任，以协调工具效能与学生能力发展，但其实际效果仍需通过真实课堂数据进一步验证。

Abstract: Cognitive offloading in human-AI collaborative teaching in a building materials course can conserve cognitive resources but may also weaken students’ independent judgment and transfer ability. Drawing on literature analysis, backward instructional design, and an “objective - task - evidence” alignment review, this study uses a coastal low-carbon concrete mix-design project to define AI permissions and students’ cognitive responsibilities. It proposes a five-stage regulatory sequence comprising “independent prediction - bounded reasoning - professional verification - scaffold removal - AI-free transfer”, together with baseline, process, and endpoint checks. Assessment indicators cover independent transfer, dependence gaps, verification quality, confidence calibration, and delayed retention. Rather than excluding AI, the framework selectively offloads secondary activities such as information retrieval and repetitive calculation while retaining students’ core cognitive responsibility for problem formulation, mechanism explanation, standards-based judgment, and engineering decisions to balance tool effectiveness with the development of students’ abilities. Its effectiveness requires further validation with authentic classroom data.