aiqbench analysis
Agentopia introduces a novel framework for long-term multi-agent simulation spanning 10 years, enabling LLMs to learn anthropomorphic social behaviors through a "life reward" mechanism. This approach successfully enhances underlying model capabilities, yielding a 15.6% improvement on downstream role-playing benchmarks while generating rich emergent social dynamics. The work marks a significant step toward evaluating LLM intelligence via extended, human-like social experiences rather than short-term interactions.
Agentopia: Long-Term Life Simulation and Learning in Agent Societies
arXiv:2606.07513v1 Announce Type: new Abstract: Humans learn from social life. Simulating this process with LLM-powered agents represents a promising research direction, raising a natural question: whether LLMs can learn from such simulated social experience to better understand and replicate human behavior. However, prior agent society simulations typically operate at the scale of days, limiting the depth of social interactions and long-term growth. In this paper, we study long-term life simulation and LLM learning in agent societies, with two goals: (1) investigating social behaviors that emerge from life-long simulation, and (2) developing anthropomorphic capabilities in LLMs, particularly intelligence in social life, through years of simulated social experience. Specifically, we present Agentopia, a comprehensive framework for long-term life simulation in multi-agent societies, where 100 agents autonomously pursue personal growth, develop social relationships, and fulfill their needs and goals over 10 simulated years. We define life reward to mirror human well-being, and leverage this reward to train LLMs via rejection sampling. Extensive experiments show that agents exhibit rich emergent social behaviors. Furthermore, life reward training effectively enhances the underlying LLM, which leads to improved agent well-being in simulation, and generalizes to downstream role-playing benchmarks with +15.6% improvement.
Abstract:Humans learn from social life. Simulating this process with LLM-powered agents represents a promising research direction, raising a natural question: whether LLMs can learn from such simulated social experience to better understand and replicate human behavior. However, prior agent society simulations typically operate at the scale of days, limiting the depth of social interactions and long-term growth. In this paper, we study long-term life simulation and LLM learning in agent societies, with two goals: (1) investigating social behaviors that emerge from life-long simulation, and (2) developing anthropomorphic capabilities in LLMs, particularly intelligence in social life, through years of simulated social experience. Specifically, we present Agentopia, a comprehensive framework for long-term life simulation in multi-agent societies, where 100 agents autonomously pursue personal growth, develop social relationships, and fulfill their needs and goals over 10 simulated years. We define life reward to mirror human well-being, and leverage this reward to train LLMs via rejection sampling. Extensive experiments show that agents exhibit rich emergent social behaviors. Furthermore, life reward training effectively enhances the underlying LLM, which leads to improved agent well-being in simulation, and generalizes to downstream role-playing benchmarks with +15.6% improvement.
Computation and Language (cs.CL)
arXiv:2606.07513 [cs.CL]
(or arXiv:2606.07513v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2606.07513 arXiv-issued DOI via DataCite (pending registration)
Full text under the publisher's open licence · 208 words
