PRISM: Personality-Driven Multi-Agent Framework for Social Media Simulation
Published:Dec 24, 2025 05:00
•1 min read
•ArXiv NLP
Analysis
This paper introduces PRISM, a novel framework for simulating social media dynamics by incorporating personality traits into agent-based models. It addresses the limitations of traditional models that often oversimplify human behavior, leading to inaccurate representations of online polarization. By using MBTI-based cognitive policies and MLLM agents, PRISM achieves better personality consistency and replicates emergent phenomena like rational suppression and affective resonance. The framework's ability to analyze complex social media ecosystems makes it a valuable tool for understanding and potentially mitigating the spread of misinformation and harmful content online. The use of data-driven priors from large-scale social media datasets enhances the realism and applicability of the simulations.
Key Takeaways
- •PRISM offers a more realistic simulation of social media dynamics by incorporating personality traits.
- •The framework uses MBTI and MLLM agents to improve personality consistency.
- •PRISM can replicate emergent phenomena like rational suppression and affective resonance.
Reference
“"PRISM achieves superior personality consistency aligned with human ground truth, significantly outperforming standard homogeneous and Big Five benchmarks."”