When Outputs Disperse, Does Epistemic Revision Follow? A Black-Box Coupling Diagnostic for Machine Collectives
作者: Molood Arman
分类: cs.AI, cs.CL
发布日期: 2026-08-04
💡 一句话要点
提出黑箱耦合诊断方法以解决集体智能中的认知修正问题
🎯 匹配领域: 支柱一:机器人控制 (Robot Control)
关键词: 集体智能 认知修正 大型语言模型 输出分散 黑箱诊断 多样性评估 实验方法 决策支持
📋 核心要点
- 现有集体智能方法在处理代理人输出分歧时,无法有效区分真正的认知修正与表面上的多样性。
- 本文提出了一种黑箱耦合诊断方法,通过分析生成文本来评估集体的认知修正能力,避免对内部模型表示的假设。
- 实验结果表明,在gpt-4o-mini配置下,条件性异议显著提高了错误前提恢复率,而在gemini-2.5-flash配置下则未见明显提升。
📝 摘要(中文)
集体智能研究将分歧视为认知多样性的证据:如果代理人表达不同的观点,群体应保留修正能力。然而,在大型语言模型(LLM)集体中,这一代理可能失效。本文通过操作化输出分散-修正耦合,评估干预是否真正导致认知立场的修正,而非仅是前提保持的重新表述。我们提出的Meta-Predictive Clarity System(MPCS)结合Re-Differentiation Protocol(RDP),用于估计这一耦合机制。实验结果显示,在不同配置的五代理集体中,条件性异议显著提高了错误前提的恢复率,而静态角色多样性则有负面影响。
🔬 方法详解
问题定义:本文旨在解决集体智能中代理人输出分歧与认知修正之间的关系,现有方法无法有效识别真正的认知修正与表面上的多样性。
核心思路:通过操作化输出分散-修正耦合,评估干预是否导致认知立场的真实修正,而非仅是前提保持的重新表述。
技术框架:整体框架包括输出通道(连贯性指数CI)和认知通道(每轮立场注释),独立测量干预对输出分散和认知修正的影响。
关键创新:提出Meta-Predictive Clarity System(MPCS)和Re-Differentiation Protocol(RDP),作为可重用的方法来估计输出分散与认知修正的耦合机制。
关键设计:在实验中,使用了310对配对实验,针对不同配置的五代理集体进行评估,重点关注条件性异议和静态角色多样性对错误前提恢复率的影响。实验结果显示,gpt-4o-mini配置下条件性异议显著提高了恢复率,而gemini-2.5-flash配置下则未见显著提升。
🖼️ 关键图片
📊 实验亮点
实验结果显示,在gpt-4o-mini配置下,条件性异议使错误前提恢复率提高了17.7个百分点(p<1e-6),而静态角色多样性则导致恢复率下降8.1个百分点(p=.007)。在gemini-2.5-flash配置下,尽管输出分散降低,但未能显著提高恢复率(26.1% vs 27.1%,p=.84),显示出不同配置间的显著差异(z=3.79, p<.001)。
🎯 应用场景
该研究的潜在应用领域包括集体智能系统、决策支持系统和多代理协作环境。通过有效评估认知修正能力,可以提升系统在复杂任务中的表现,增强其适应性和灵活性。未来,该方法有望在更广泛的AI应用中得到推广,促进智能体之间的有效协作。
📄 摘要(原文)
Collective intelligence research treats disagreement as evidence of epistemic diversity: if agents express different views, the group should retain capacity to revise. In LLM collectives this proxy can break: agents can produce diverse-looking arguments while preserving the same conclusion. We operationalize dispersion-revision coupling: the degree to which an intervention that verifiably increases the dispersion of a collective's outputs in embedding space is accompanied by genuine revision of its epistemic stance rather than premise-preserving reformulation. The diagnostic is black-box: it operates on generated text alone and makes no claims about the internal representations of the generating models. Two channels are measured independently: an output channel, the Coherence Index (CI), verifies that the intervention changed output dispersion; an epistemic channel, per-turn stance annotation, measures whether the collective revised. We propose CI with the Meta-Predictive Clarity System (MPCS), which inserts a Re-Differentiation Protocol (RDP) when outputs over-converge, as a reusable method for estimating this coupling regime. We evaluate five-agent collectives from two configurations (gpt-4o-mini and gemini-2.5-flash; 310 paired episodes per condition). On gpt-4o-mini, conditional dissent improves false-premise recovery by +17.7 points (p<1e-6) while static persona diversity harms recovery (-8.1, p=.007). On gemini-2.5-flash, the same intervention at a comparable budget yields no gain (26.1% vs 27.1%, p=.84) despite a verified dispersion drop; the two treatment effects differ from each other (z=3.79, p<.001). Mechanism tagging shows Gemini preserves the false premise via intra-framework dissent: 94% of tagged post-RDP responses reformulate rather than concede (vs 24% on GPT). We recommend reporting per-intervention stance shift and premise-preservation rate alongside accuracy.