Generative Agents

Interactive Simulacra of Human Behavior

Bibliographic Data

ID23344764
AuthorsJoon Sung Park (0000-0001-5036-4409, Stanford University), Joseph O’Brien (0009-0004-0781-926X, Stanford University), Carrie Jun Cai (0000-0001-9421-7128, Google (United States)), Meredith Ringel Morris (0000-0003-1436-9223, Google (United States)), Pinghan Liang (0000-0002-0458-6139, Stanford University), Michael S Bernstein (0000-0001-8020-9434, Stanford University)
Year2023
Pages1-22
Publication date2023-10-29
Peer ReviewedYes
Open AccessYes
TypeCONFERENCE
VenueProceedings of the 36th Annual ACM Symposium on User Interface Software and Technology (CONFERENCE)
PublisherACM (PUBLISHER)
DOI10.1145/3586183.3606763
OpenAlexW4387835442
LanguageEN
Citations received92
References cited50

Believable proxies of human behavior can empower interactive applications ranging from immersive environments to rehearsal spaces for interpersonal communication to prototyping tools. In this paper, we introduce generative agents: computational software agents that simulate believable human behavior. Generative agents wake up, cook breakfast, and head to work; artists paint, while authors write; they form opinions, notice each other, and initiate conversations; they remember and reflect on days past as they plan the next day. To enable generative agents, we describe an architecture that extends a large language model to store a complete record of the agent’s experiences using natural language, synthesize those memories over time into higher-level reflections, and retrieve them dynamically to plan behavior. We instantiate generative agents to populate an interactive sandbox environment inspired by The Sims, where end users can interact with a small town of twenty-five agents using natural language. In an evaluation, these generative agents produce believable individual and emergent social behaviors. For example, starting with only a single user-specified notion that one agent wants to throw a Valentine’s Day party, the agents autonomously spread invitations to the party over the next two days, make new acquaintances, ask each other out on dates to the party, and coordinate to show up for the party together at the right time. We demonstrate through ablation that the components of our agent architecture—observation, planning, and reflection—each contribute critically to the believability of agent behavior. By fusing large language models with computational interactive agents, this work introduces architectural and interaction patterns for enabling believable simulations of human behavior.

Generative grammar · Generative model · Human–computer interaction · Artificial Intelligence · Artificial Intelligence in Games · Computer Science · Reinforcement Learning in Robotics · Social Robot Interaction and HRI

Tools

Open DOIOpen Access