arXiv：cs.AI· Dimitris Michailidis, Petter T\"ornberg, Fernando P. Santos·· 6 小时前
从漫画化到集体产物：人格构建如何驱动生成式社会模拟中的极化
From Caricature to Collective Artifact: How Persona Construction Drives Polarization in Generative Social Simulations
AI 导读
研究基于 Twitter 式生成式社会模拟平台，用美国国家选举研究（ANES）数据构建 LLM 智能体人格，剥离情感极化标记、党派认同和投票选择三层党派身份。结果显示，仅移除少数明确党派表述，网络就从近乎完全隔离转向接近真实 Twitter 的跨党派连接。作者认为单一结果匹配不足以证明行为有效性，并指出问卷中不重要的议题回答被转为人格指令后，其显著性可能被过度放大。
正文
Abstract:Generative Social Simulation (GSS) leverages Large Language Models (LLMs) to create sophisticated personas that define agents with demographics, emotions, and political opinions. While prior work shows that personas compress outputs into stereotypical caricatures, there is no comprehensive study of how these caricatures evolve into collective artifacts when they interact. Using a Twitter-like GSS platform with personas derived from the American National Elections Study (ANES), we ablate three layers of partisan identity: affective polarization markers, party identification, and vote choice, while retaining demographics, ideological positioning and important issues. We find that collective outcomes depend heavily on how survey responses are translated into personas. Removing a few explicitly partisan statements moves the network from near-complete segregation to cross-party ties approaching those of real Twitter networks. We argue that matching one outcome is not evidence for behavioral validity and point to salience as a possible explanation. Surveys ask for a position on every question regardless of its importance to the respondent, and persona construction turns each answer into a direct behavioral instruction. We conclude with a checklist for more comprehensive evaluation of personas in future works.
| Comments: | 17 pages |
| Subjects: | Social and Information Networks (cs.SI); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.16498 [cs.SI] |
| (or arXiv:2610.16498v1 [cs.SI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.16498 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Dimitris Michailidis [view email]
[v1]
Mon, 12 Oct 2026 09:48:14 UTC (217 KB)
来源：arXiv：cs.AI · arxiv.org