The convergence of artificial intelligence and complex system modeling has opened new opportunities for exploring swarm intelligence and emergent behaviors. Traditional multi-agent simulations rely on predefined rule sets, limiting adaptability. By integrating Large Language Models (LLMs), we shift from rigid systems to dynamic, knowledge-driven interactions. This work leverages LLMs to guide agent behavior in realtime, offering insights into self-organization and collective intelligence. Using the NetLogo simulation environment, we explore ant colony foraging and bird flocking, demonstrating how LLMs can facilitate emergent behaviors inspired by natural phenomena.

 Setup: The simulation models ants foraging for food in a twodimensional environment using structured prompts generated by ChatGPT-4.

 LLM Integration: ChatGPT-4 processes environment feedback to dictate ant behaviors like moving forward, picking up food, and following pheromone trails.

 Prompt Design: Zero-shot prompts ensure each agent response is stateless, relying solely on real-time inputs.