Reproduce “Y Social: an LLM-powered Social Media Digital Twin” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Giulio Rossetti, Massimo Stella, Rémy Cazabet, Katherine Abramski, Erica Cau, Salvatore Citraro, Andrea Failla, Riccardo Improta and 2 more
Researchers developed Y Social, a digital twin of social media platforms powered by Large Language Models, enabling the simulation and experimental investigation of platform dynamics, algorithmic effects, and user interactions in a controlled environment. The platform generates realistic user behaviors and content, mirroring patterns observed in real-world social media.
Key insights
- Large Language Models can effectively power agents in a digital twin to simulate complex and realistic social media behaviors and content generation.
- A controlled digital twin environment allows for systematic experimental investigation of social media platform dynamics and the impact of different algorithmic curation strategies.
- Simulated interactions within such a system can reproduce observed patterns in real-world online discourse, including content distribution, emotional expressions, and political engagement.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.