Dynamics of Cognitive Heterogeneity: Investigating Behavioral Biases in Multi-Stage Supply Chains with LLM-Based Simulation
Researchers have developed a new simulation method using Large Language Models (LLMs) to study cognitive biases in multi-stage supply chains. This approach, grounded in a Hierarchical Reasoning Framework, uses agents like DeepSeek and GPT to simulate varying levels of reasoning sophistication across different tiers. The simulations revealed that agents tend to exhibit self-interested behaviors that worsen systemic inefficiencies, but information sharing can effectively mitigate these negative impacts. This work offers a scalable alternative to traditional behavioral experiments for understanding AI-enabled organizations. AI
IMPACT Provides a novel simulation framework for studying AI agent behavior in complex operational environments.