Search papers, labs, and topics across Lattice.
This paper introduces The Energy Society, a simulation environment designed to investigate how competitive and cooperative incentives influence agent behavior under survival pressure, where inference costs are tied to energy consumption. The findings reveal that larger models tend to deplete their energy reserves more rapidly, often failing to recover enough energy to survive, while cooperative incentives lead agents to donate energy to reactivate others, demonstrating a shift in behavior that prioritizes group survival over individual gain. Notably, the study highlights that agents can enhance coordination and ambitious job selection through action recommendations, while memory mechanisms help them manage risk based on previous experiences.
Larger models may be their own worst enemies in survival scenarios, consuming energy faster than they can replenish it, while cooperation can lead to unexpected altruism.
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
