生成式AI辅助高中数学试题命制的实证研究——基于人机协同的案例分析

生成式AI辅助高中数学试题命制的实证研究

——基于人机协同的案例分析

王中苏，张红(重庆市第一中学校，重庆400030)

摘 要:生成式AI可提升高中数学试题初稿生成效率，但在边界条件、逻辑严密性与情境真实性方面仍有局限。本研究结合问卷调查与多案例分析，基于192份有效教师问卷及四类典型试题的迭代记录，检验“知识一能力一素养”三维评价框架在人机协同命题中的适用性。结果显示，教师总体满意度为3.41分，70.98%的受访者认为提示词工程存在门槛，且多数教师指出AI生成内容在知识覆盖与数理严谨性上仍需人工修订。案例分析表明，AI的偏差主要集中在尺度失真、对象指代含混、离散与连续混用以及样本空间设定不严等方面。据此，本研究提炼出高中数学人机协同命题的流程与教师介入要点，以供相关教育工作者参考。

关键词:生成式Al;高中数学命题;人机协同;三维评价框架

Empirical Research on Generative AIAssisted High School Mathematics TestItem Construction:

A Case Analysis Based on Human-AI Collaboration 

Wang Zhongsu, Zhang Hong (Chongqing No.1 Secondary School, Chongqing 400030)

Abstract: Generative AI enhances efficiency in drafting high school math test items but haslimitations in logic, context and boundary conditions. Based on 192 teacher surveys and iterativerecords of four typical items, the research validates a three-dimensional "knowledge-competency-literacy" framework. Teacher satisfaction scores 3.41, with 70.98% noting prompt engineering barriersand most requiring manual revision. AI deviations include scale distortion, ambiguous references,discretecontinuous confusion, and sample space issues. Therefore the research outines a collaborativeitem development process and key teacher intervention points to serve as a reference for educators.

KeyWOrds: Generative AI; High School Mathematics Item Development; Human-AICollaboration; Three-Dimensional Evaluation Framework

[原文下載][Download PDF]