arXiv cs.AI· Chaemin Jang, Junsik Min, Jaewoo Choi, Donggyu Lee, Haiin Lee, Yoonseo Choi, Junyoung Park, Namhee Kim, Hyunwoo Kim, Juho Kim, Nuri Kim, Jihee Kim·· 2 天前AI 评分31
从模拟公民到模拟审议：LLM 智能体在代表性与交互上的挑战
From Simulated Citizens to Simulated Deliberation: Challenges in Representation and Interaction
AI 导读
研究用全国调查、受控审议实验和真实审议民调评估六款 LLM 能否模拟公众审议，发现智能体常产生过度集中的意见，无法可靠复现人口统计差异。在模拟审议中智能体虽会交换论点并频繁改变立场，但在看不到彼此发言时也会出现类似变化，而仅重述初始立场则大幅减少立场变动。
正文 · 原文
Authors:Chaemin Jang, Junsik Min, Jaewoo Choi, Donggyu Lee, Haiin Lee, Yoonseo Choi, Junyoung Park, Namhee Kim, Hyunwoo Kim, Juho Kim, Nuri Kim, Jihee Kim
Abstract:Can LLM agents simulate how a real public would deliberate? For such simulations to be meaningful, agents must reflect the population's opinions, and their views must be shaped by what others say during discussion. We evaluate these requirements across six LLMs using national surveys, controlled deliberation experiments, and a real deliberative poll. Comparing census-grounded Korean personas against national surveys, we find that agents often produce overly concentrated opinions and fail to reliably reproduce demographic differences. During simulated deliberation, agents exchange arguments, respond to one another, and frequently change positions. Yet similar changes occur even when agents cannot see one another's contributions, while simply restating their starting positions sharply reduces movement. We further test whether these limitations persist in a real-world setting, using agents constructed from actual respondents in the 2017 Shin-Kori deliberative poll. Even with real respondent profiles, agents fail to reproduce the panel's initial opinion distribution or how its opinions change across survey waves. Our findings show that generating deliberative discussions, representing public opinion, and reproducing interaction-driven opinion change are distinct capabilities. These results suggest a different potential role for LLM deliberation: surfacing arguments and perspectives, rather than simulating a real public.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2609.07573 [cs.AI] |
| (or arXiv:2609.07573v2 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2609.07573 arXiv-issued DOI via DataCite |
Submission history
From: Chaemin Jang [view email]
[v1]
Mon, 7 Sep 2026 14:49:25 UTC (710 KB)
[v2]
Thu, 15 Oct 2026 03:39:35 UTC (196 KB)
来源：arXiv cs.AI · arxiv.org