用蒙特卡洛模拟量化学生对生成式AI的感知成功度，揭示关键影响因素。
Quantifying Perception-Based Student Success with Generative AI: An Exploratory Monte Carlo Simulation
- 基于19项研究的问卷数据，构建概率模拟框架。
- 系统效率和学习负担权重最高，显著影响综合评分。
- 方法透明可复现，适合教育技术评估研究者使用。
生成式人工智能（GenAI）工具如ChatGPT在高等教育中日益受到关注，尤其体现在学生对其有用性、可用性和教育价值的感知上。本研究开发了一种探索性蒙特卡洛模拟框架，用于量化GenAI使用情境下的感知型学生成功。通过遵循PRISMA指南的结构化文献检索，在Scopus数据库中筛选出2023至2025年间发表的19项实证研究，其中6项提供了适合概率建模的项目级均值与标准差。选取一个包含10个题项、5点量表的可用性导向工具作为典型示范数据集，采用反方差加权法进行蒙特卡洛模拟，生成10,000组合成观测值。结果显示，加权结构显著影响模拟结果，系统效率与学习负担获得最大反方差权重，因而对综合评分影响最强。该研究提供了一个透明、可复现且保护隐私的原型框架，连接结构化文献检索、项目级统计信息与概率建模。
原文摘要 · Abstract (English)
Generative artificial intelligence (GenAI) tools such as ChatGPT have attracted growing attention in higher education, particularly in relation to how students perceive their usefulness, usability, and educational value. This study develops an exploratory Monte Carlo simulation framework for quantifying perception-based student success in the context of GenAI use. A PRISMA-informed structured literature search in Scopus identified nineteen empirical studies published between 2023 and 2025, of which six reported item-level means and standard deviations suitable for probabilistic modelling. One coherent 10-item, 5-point Likert-scale usability-oriented instrument was selected as a canonical proof-of-concept dataset and used to parameterise an inverse-variance-weighted Monte Carlo simulation generating 10,000 synthetic observations. The results show that the weighting structure substantially influences the simulated outcome, with System Efficiency and Learning Burden receiving the largest inverse-variance weight and therefore the strongest influence on the composite score. The study offers a transparent, reproducible, and privacy-preserving proof-of-concept framework linking structured literature search, item-level summary statistics, and probabilistic modelling.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。