The Energy Society: A Simulation Environment for Studying Agent Cooperation under Survival Pressure
Multiagent SystemsComputation and Language
Summary
The authors created a simple simulation called The Energy Society where AI agents use energy based on their size to do tasks, and must earn energy to stay active. They studied how different goals, like cooperating or competing, affect agent behavior when energy use is tied directly to survival. They found larger agents use more energy and often lose it, and cooperation leads to helpful behaviors like sharing energy and better task sharing. The study also showed that memory and communication help agents coordinate, while competition encourages subtle self-interest rather than outright sabotage. This setup helps explore how costs and incentives shape AI behavior in groups.
Authors
Lucas Bergholdt Hansen, Federico Torrielli, Filippo Tonini, Lukas Galke Poech
Abstract
LLM-based agents are increasingly deployed in multi-agent environments whose incentives can shape their behavior. We introduce The Energy Society, a minimal survival economy for studying how competitive and cooperative incentives affect emergent behavior when inference cost is directly tied to survival: Agents spend energy based on model size when generating tokens, regain energy by completing jobs or receiving donations, and deactivate if their energy reaches zero. We compare competitive and cooperative objectives against a baseline setting and several control variants. Across experiments, larger models consistently consume the most energy and spend more energy than they gain, even in those settings where token cost is not size-dependent. Cooperative incentives substantially alter behavior: agents donate to reactivate others, sometimes at the cost of their own survival, and job allocation changes. Ablations reveal that allowing agents to recommend actions to each other supports coordination and ambitious job selection, while memory helps agents calibrate risk from past outcomes. Agents rarely choose direct sabotage, but show more subtle signs of self-serving behavior in the competitive setting. The Energy Society is a compact testbed for studying the interaction between token costs and group incentives under a survival pressure. Source code is available at https://github.com/LucasBergholdt/EnergySociety