camel-ai/oasis

What it solves

OASIS is designed to study complex social phenomena in digital environments, such as information spread, group polarization, and herd behavior. It provides a scalable way to mimic the behavior of large populations of users on social media platforms like Twitter and Reddit without needing real-world data for every interaction.

How it works

The system uses large language model (LLM) agents to simulate users. It can scale up to one million agents, each capable of performing 23 different actions (e.g., following, commenting, reposting). The simulation environment incorporates interest-based and hot-score-based recommendation algorithms to mirror how content is discovered and distributed on real social networks.

Who it’s for

This tool is primarily for researchers studying social dynamics, digital sociology, and the impact of AI agents on social media interactions.

Highlights

• Massive Scalability: Supports simulations of up to one million agents.
• Dynamic Environments: Mimics the fluid nature of real-time changes in social networks and content.
• Rich Action Space: Agents can perform 23 distinct actions to create multi-faceted interactions.
• Integrated Recommendations: Includes built-in recommendation systems to simulate realistic content discovery.