Scholarlink

<p>This study pioneers the integration of large language models (LLMs) into simulation-based inventory optimization under supply chain disruptions. We develop and compare three human–AI collaborative approaches with varying degrees of automation: fully automated direct optimization (FADO), two-stage automated simulation and optimization (TASO), and simulation-based human-specified optimization (SHSO). Through extensive numerical experiments, we evaluate their performance using the probability of correct selection and solution quality under different operational conditions. Results show that while FADO offers the greatest automation and ease of implementation, its accuracy and stability are limited. TASO enhances computational robustness but remains sensitive to the granularity of parameters. SHSO consistently achieves statistically optimal and stable outcomes by combining LLM-driven simulation with human-specified algorithms. These findings advance the theoretical understanding of human–AI collaboration in operations and provide actionable guidance for managers seeking to enhance decision reliability and supply chain resilience.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Inventory optimization under supply chain disruptions: Leveraging large language models for human-AI collaborative decision-making

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月19日

查看全文
  • Ruijing Wu

摘要

This study pioneers the integration of large language models (LLMs) into simulation-based inventory optimization under supply chain disruptions. We develop and compare three human–AI collaborative approaches with varying degrees of automation: fully automated direct optimization (FADO), two-stage automated simulation and optimization (TASO), and simulation-based human-specified optimization (SHSO). Through extensive numerical experiments, we evaluate their performance using the probability of correct selection and solution quality under different operational conditions. Results show that while FADO offers the greatest automation and ease of implementation, its accuracy and stability are limited. TASO enhances computational robustness but remains sensitive to the granularity of parameters. SHSO consistently achieves statistically optimal and stable outcomes by combining LLM-driven simulation with human-specified algorithms. These findings advance the theoretical understanding of human–AI collaboration in operations and provide actionable guidance for managers seeking to enhance decision reliability and supply chain resilience.