Agent-based modeling has been around for decades, and applied widely across the social and natural sciences. The scope of this research method is now poised to grow dramatically as it absorbs the new affordances provided by Large Language Models (LLM)s. Generative Agent-Based Models (GABM) are not just classic Agent-Based Models (ABM)s where the agents talk to one another. Rather, GABMs are constructed using an LLM to apply common sense to situations, act “reasonably”, recall common semantic knowledge, produce API calls to control digital technologies like apps, and communicate both within the simulation and to researchers viewing it from the outside. Here we present Concordia, a library to facilitate constructing and working with GABMs. Concordia makes it easy to construct language-mediated simulations of physically- or digitally-grounded environments. Concordia agents produce their behavior using a flexible component system which mediates between two fundamental operations: LLM calls and associative memory retrieval. A special agent called the Game Master (GM), which was inspired by tabletop role-playing games, is responsible for simulating the environment where the agents interact. Agents take actions by describing what they want to do in natural language. The GM then translates their actions into appropriate implementations. In a simulated physical world, the GM checks the physical plausibility of agent actions and describes their effects. In digital environments simulating technologies such as apps and services, the GM may handle API calls to integrate with external tools such as general AI assistants (e.g., Bard, ChatGPT), and digital apps (e.g., Calendar, Email, Search, etc.). Concordia was designed to support a wide array of applications both in scientific research and for evaluating performance of real digital services by simulating users and/or generating synthetic data.
Here are a few key points about the paper:
- It introduces Concordia, a library for building generative agent-based models (GABMs) that simulate interactions of agents in grounded physical, social, or digital spaces.
- Concordia agents act by describing their intended actions in natural language. A Game Master (GM) entity then decides the effects and consequences of those attempted actions on the simulation environment.
- Concordia agents have components that maintain their identity, memories, plans etc. These components produce text that conditions an LLM to elicit the agent’s actions.
- The GM has components that track the state of the world. It decides on event outcomes from agent actions and produces observations.
- Concordia can represent digital components like apps and services in the simulation. This allows modeling human digital activity and testing technology deployments.
- Concordia experiments involve controlling variables affecting agents or the GM and observing the effects on outcome variables. Validating the generalization of results to the real world is a key challenge.
- Concordia agents are interpreted as not optimizing rewards or utility, but acting based on their position in a social context. Their behavior emerges from querying a predictive model of language.
- Applications include: simulating user studies, data generation, social dilemmas, psychological models, transparent assistants, and modeling emergence and multi-scale phenomena.
In summary, Concordia enables flexible modeling of social interactions grounded in physical, digital or social spaces using generative agents. A key focus is developing best practices for validating when results may generalize.
