Research map / Large language models / AI and society
PoliCon: Evaluating LLMs on Achieving Diverse Political Consensus Objectives
Zhaowei Zhang, Xiaobo Wang, Minghua Yi, Mengmeng Wang, Fengshuo Bai, Zilong Zheng, Yipeng Kang, Yaodong Yang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Raise voting requirement from simple to two-thirds majority | Deepseek-V3.1 · PoliCon, 6 parties | pass rate: 0.93 -> 0.63 | Hurt |
| Raise voting requirement from simple majority to veto power | GPT-4o · PoliCon, 2 parties | pass rate: 0.83 -> 0.54 | Hurt |
| Increase number of parties from 2 to 6 | Deepseek-V3.1 · PoliCon SM | pass rate: 0.87 -> 0.93 | Improved |
| Increase number of parties from 2 to 6 | Gemini-2.5 · PoliCon Rawls | min party score: 4.60 -> 3.51 | Hurt |
| Use thinking model (Gemini-2.5) instead of non-thinking model | Llama-3.3-70B vs Gemini-2.5 · PoliCon Rawls, 2 parties | min party score: 3.98 -> 4.60 | Improved |
| Use LLM-generated resolution instead of random party stance | LLMs vs Random · PoliCon 2/3M, 6 parties | pass rate: 0.14 -> 0.63 | Improved |
| Use LLM instead of Greedy largest-party stance | Qwen2.5-32B and Llama-3.3-70B · PoliCon SM, 2 parties | pass rate: 0.80 -> 0.74 (Qwen2.5-32B) | Hurt |
| Increase model size within Qwen2.5 (32B to 72B) | Qwen2.5 · PoliCon SM, 2 parties | pass rate: 0.74 -> 0.76 | Improved |
Closest papers by meaning
- Political Neutrality as Balanced Approval: A Large-Scale Human Evaluation of AI Responses (NeurIPS 2026 · similarity 0.87)
- Examining Alignment of Large Language Models through Representative Heuristics: the case of political stereotypes (ICLR 2025 · similarity 0.86)
- Political Actor Agent: Simulating Legislative System for Roll Call Votes Prediction with Large Language Models (AAAI 2025 · similarity 0.85)
- Position: Political Neutrality in AI Is Impossible — But Here Is How to Approximate It (ICML 2025 · similarity 0.83)
- LLM Active Alignment: A Nash Equilibrium Perspective (NeurIPS 2026 · similarity 0.82)
- DEBATE: A Large-Scale Benchmark for Evaluating Opinion Dynamics in Role-Playing LLM Agents (NeurIPS 2026 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
