Paper Reading Journals #9
Generative Agents: Interactive Simulacra of Human Behavior
Basic Info
| Item | Content |
|---|---|
| Title | Generative Agents: Interactive Simulacra of Human Behavior |
| Journal | Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology |
| Author | Joon Sung Park, Joseph C. O’Brien, Carrie J. Cai, Meredith Ringel Morris, Percy Liang, Michael S. Bernstein |
| Year Published | 2023 |
| Link | https://doi.org/10.1145/3586183.3606763 |
| Geography of Study | N/A |
| Area of Study | Generative Agent, Simulation |
Description
(Including data and methods)
Use GPT3.5-Turbo to generate synthetic agent who has memory and were able to think and plan. In order to simulate Real-World interaction. Data: Generate by GPT3.6-Turbo Method: Agent-Based Simulation (ABM), Social Simulation, dynamic memory retrieval.
Main Contribution
(Innovative method or Application) Innovation Type: Method Innovation This work innovative adopt LLM powered generative AI agent in the simulation of social interactions. It propose and valid the potential of utilize generative AI agent in simulating complex interaction between society.
Data Source
Synthetic agent generated by LLM GPT-3.5 Turbo. Validation: 100 Human Evaluator
Favorite Thing
The concept, and especially how they visualize and present their work is interesting. In addition, they provide a basic framework and platform of performing AI-LLM driven agent based simulation which we could adopt to simulate healthcare system interactions.