用多模型对话测试AI对齐策略，发现不同模型各有侧重。
Dialogical Reasoning Across AI Architectures: A Multi-Model Framework for Testing AI Alignment Strategies
- 用四个角色分工让不同AI模型对话，模拟对齐问题讨论。
- 72轮对话中，模型提出互补意见并生成新见解，如VCW过渡框架。
- 适合研究AI对齐、人机协作及和平学方法的学者参考。
本文提出一种通过结构化多模型对话实证测试AI对齐策略的方法框架。借鉴和平研究中的利益协商、冲突转化与公共治理理念，将对齐重构为通过对话推理建立关系的问题，操作化实现Viral Collaborative Wisdom（VCW）方法。实验设计在六种条件下，由Claude、Gemini和GPT-4o担任四类角色（提案者、回应者、监督者、翻译者），完成72轮对话，共576,822字符的结构化交流。结果表明，大模型能有意义地参与和平学概念，从不同架构视角揭示互补性质疑，并生成初始框架中未包含的新兴洞察，包括“VCW作为过渡框架”的新合成。跨架构模式显示：Claude关注验证挑战，Gemini聚焦偏见与可扩展性，GPT-4o强调实施障碍。该框架为研究者提供可复现的对齐提案压力测试方法，研究结果初步证明了大模型具备对话推理能力。论文也指出局限，如对话更关注过程而非对AI本质的深层主张，并建议未来探索人机混合协议与延长对话研究。
原文摘要 · Abstract (English)
This paper introduces a methodological framework for empirically testing AI alignment strategies through structured multi-model dialogue. Drawing on Peace Studies traditions - particularly interest-based negotiation, conflict transformation, and commons governance - we operationalize Viral Collaborative Wisdom (VCW), an approach that reframes alignment from a control problem to a relationship problem developed through dialogical reasoning. Our experimental design assigns four distinct roles (Proposer, Responder, Monitor, Translator) to different AI systems across six conditions, testing whether current large language models can engage substantively with complex alignment frameworks. Using Claude, Gemini, and GPT-4o, we conducted 72 dialogue turns totaling 576,822 characters of structured exchange. Results demonstrate that AI systems can engage meaningfully with Peace Studies concepts, surface complementary objections from different architectural perspectives, and generate emergent insights not present in initial framings - including the novel synthesis of "VCW as transitional framework." Cross-architecture patterns reveal that different models foreground different concerns: Claude emphasized verification challenges, Gemini focused on bias and scalability, and GPT-4o highlighted implementation barriers. The framework provides researchers with replicable methods for stress-testing alignment proposals before implementation, while the findings offer preliminary evidence about AI capacity for the kind of dialogical reasoning VCW proposes. We discuss limitations, including the observation that dialogues engaged more with process elements than with foundational claims about AI nature, and outline directions for future research including human-AI hybrid protocols and extended dialogue studies.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。