Scholarlink

Artificial Intelligence (AI) agents are increasingly embedded in collaborative supply chains, introducing not only opportunities for efficiency but also systemic risks. This paper develops a systems-based framework for evaluating AI-agent risks by integrating DEMATEL and System Dynamics. The study identifies eight interdependent risk categories, including autonomy overreach, trust erosion, ethical misalignment, and governance gaps. Through expert input and simulation modeling, we demonstrate how certain risks function as causal drivers, while others act as amplifiers in cascading effects. Results show that autonomy overreach and ethical misalignment are the most influential risks, while proactive governance mechanisms significantly mitigate their impact. Contributions of this paper are threefold: (i) mapping AI-agent risks within collaborative supply chains, (ii) developing a hybrid DEMATEL–System Dynamics framework for systemic risk evaluation, and (iii) simulating governance strategies to inform resilient policy and decision-making.

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

A Systems Approach to Evaluating AI Agent Risks in AI-Based Planning and Collaborative Supply Chains

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Javad Jassbi,
  • Roozbeh Aliabadi,
  • Jose Barata

摘要

Artificial Intelligence (AI) agents are increasingly embedded in collaborative supply chains, introducing not only opportunities for efficiency but also systemic risks. This paper develops a systems-based framework for evaluating AI-agent risks by integrating DEMATEL and System Dynamics. The study identifies eight interdependent risk categories, including autonomy overreach, trust erosion, ethical misalignment, and governance gaps. Through expert input and simulation modeling, we demonstrate how certain risks function as causal drivers, while others act as amplifiers in cascading effects. Results show that autonomy overreach and ethical misalignment are the most influential risks, while proactive governance mechanisms significantly mitigate their impact. Contributions of this paper are threefold: (i) mapping AI-agent risks within collaborative supply chains, (ii) developing a hybrid DEMATEL–System Dynamics framework for systemic risk evaluation, and (iii) simulating governance strategies to inform resilient policy and decision-making.