基于MTA-TPACK动态协作螺旋的AI智能体辅助教学实践——以热带气旋WISHE机制科学图像生成为例

AI Agent for Teaching Practice Grounded in the MTA-TPACK Dynamic Collaboration Spiral—A Case Study of Scientific Image Generation for the Tropical Cyclone WISHE Mechanism

摘要: 如何使教师的学科知识在人机协作中转化为可直接应用于课堂教学的科学准确的视觉表征，是生成式AI赋能科学教育教学实践的核心挑战。本研究以MTA-TPACK动态协作螺旋为分析框架，将其所刻画的“知识激活–诊断评价–后设修正”过程逻辑编码为基于Gemini平台的AI智能体(GEM)，以支持教师在协作过程中进行判断、选择与调整，引导教师完成从文献溯源到视觉约束转化、从量规诊断到迭代精修的闭环教学设计协作。以热带气旋风致表面热交换(WISHE)机制为案例，围绕二次眼墙形成(SEF)与快速增强(RI)展开科学图像的螺旋式优化。结果显示，SEF图像科学准确性由4分跃升至满分，RI由4.7分提升至4.9分。智能体自动输出的可见教学思维(VPT)轨迹，将隐性学科判断凝练为可追溯的教学资源，记录了从结构描述到机制解释再到物理完备化的认知跃迁。本研究为以MTA-TPACK为分析视角设计可操作的课堂教学辅助智能工具提供了方法论路径。

Abstract: Transforming teachers’ disciplinary knowledge into scientifically accurate visual representations through human-AI collaboration remains a central challenge in applying generative AI to science education teaching practice. Grounded in the MTA-TPACK Dynamic Collaboration Spiral, this study encodes its three-stage process logic—knowledge activation, diagnostic evaluation, and meta-reflective refinement—into an AI agent (GEM) built on the Google Gemini platform, guiding teachers through a closed-loop workflow from literature-grounded constraint translation to rubric-based iterative refinement within an instructional design workflow. Using the Wind-Induced Surface Heat Exchange (WISHE) mechanism of tropical cyclones as an empirical case, spiral-based optimization is conducted across Secondary Eyewall Formation (SEF) and Rapid Intensification (RI). Results show that SEF scientific accuracy rose from 4 to a perfect 5 out of 5, while RI improved from 4.7 to 4.9. The Visible Pedagogical Thinking (VPT) trajectory automatically generated by the agent distills tacit disciplinary judgments into traceable pedagogical resources, documenting a cognitive progression from structural description through mechanistic explanation to physical completeness. This study provides a methodological pathway for designing operational, classroom-ready intelligent tools informed by the MTA-TPACK framework.