发表机构
MetaAI（MetaAI）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究揭示LLM对话失败源于早期后验坍缩，即过早承诺模糊意图，而非记忆遗忘；提出应保留不确定性、暂持假设并在高影响模糊时询问，以提升鲁棒性。
AI 中文摘要
语言模型中的对话失败通常被归结为记忆失败：上下文过长、摘要丢失信息、约束被遗忘。我们认为这忽略了一个更深层的问题：在许多对话中，模型并非遗忘，而是过早地做出承诺。一个模糊的早期轮次会坍缩为单一的隐藏解释，后续的澄清也会被这种承诺所过滤。我们称此为早期后验坍缩：在模糊性得到解决之前，未解决的用户意图坍缩为已承诺的任务状态。我们使用Gemini-2.5-Pro和Gemini-2.5-Flash，在写作、规划和编码的受控对话任务中对此进行研究。在数千次试验中，相同的信息以不同的顺序呈现会产生不同的结果，即使最终对话包含等效的任务相关信息。这种顺序效应表明，后续的澄清被视为额外的上下文而非纠正信号：它细化了一个过时的任务状态，却没有使其失效。编码任务尤其容易受到影响，这表明早期假设会嵌入到接口和控制流等结构化工件中。标准的提示和记忆策略并不能可靠地提供帮助：摘要可能坍缩模糊性，而思维链可以减少推理轨迹中明确的错误承诺，却无法提高最终任务的成功率。这些发现促使我们采用保留不确定性的状态管理。如果助手无法放弃早期的解释，鲁棒性就不能仅依赖于事后纠正；它必须防止模糊的早期轮次固化为单一的任务状态。在模糊性仍然存在时，助手应持有暂定假设；当高影响的模糊性持续存在时，应在执行前询问；当后续证据使早期解读失效时，应从修订后的状态重建。我们并非提供单一的提示修复方案，而是旨在将交互式LLM的研究从保留更多上下文转向保留不确定性。
英文摘要
Dialogue failures in language models are usually framed as memory failures: context too long, summaries lossy, a constraint forgotten. We argue this misses a deeper problem: in many conversations the model does not forget, it commits too early. An ambiguous early turn collapses into a single hidden interpretation, and later clarification is filtered through that commitment. We call this early posterior collapse: unresolved user intent collapsing into a committed task state before ambiguity is resolved. We study it with controlled dialogue tasks in writing, planning, and coding using Gemini-2.5-Pro and Gemini-2.5-Flash. Across thousands of trials, the same information in different orders yields different outcomes, even when the final dialogue contains equivalent task-relevant information. This order effect suggests later clarification is treated as extra context rather than a corrective signal: it refines a stale task state without invalidating it. Coding tasks are especially vulnerable, suggesting early assumptions get embedded in structured artifacts such as interfaces and control flow. Standard prompting and memory strategies do not reliably help: summaries can collapse ambiguity, and chain-of-thought can reduce explicit wrong commitment in reasoning traces without improving final task success. These findings motivate uncertainty-preserving state management. If assistants cannot let go of early interpretations, robustness cannot rely on post hoc correction alone; it must keep ambiguous early turns from hardening into one task state. Assistants should hold tentative hypotheses while ambiguity remains, ask before executing when high-impact ambiguity persists, and rebuild from a revised state when later evidence invalidates an earlier reading. Rather than one prompting fix, we aim to redirect research for interactive LLMs from retaining more context toward preserving uncertainty.
Comments17 pages
