Position: AI Agents Are Not (Yet) a Panacea for Social Simulation

Researchers argue that LLM-based AI agents are not yet effective for social simulation, despite growing optimism in the field. The paper identifies systematic mismatches between what current agent systems produce and what scientific simulation requires, calling for more rigorous validation frameworks.

• →Current LLM-based agents fail to meet the requirements for scientifically valid social simulation despite recent advances.
• →Role-playing plausibility in AI agents does not guarantee faithful representation of human behavioral patterns.
• →Collective simulation outcomes are heavily influenced by interaction protocols and scheduling rather than just agent interactions.
• →The paper proposes a unified mathematical framework for AI agent-based social simulation using Markov game theory.
• →Researchers call for explicit validation mechanisms to make simulation assumptions auditable and transparent.