语言模型生成中的接地缺口
Grounding Gaps in Language Model Generations
AI总结：
本文构建接地行为与量化指标，对比多对话数据集中的LLM与人类发言，发现LLM更少进行澄清和确认，并揭示偏好优化训练会加剧这一接地缺口。
摘要：
有效的对话需要共同基础：参与者之间的共享理解。然而，共同基础并不会在对话中自发出现。说话者和倾听者需要协同努力，以识别并构建共享基础，同时避免误解。为了完成接地，人类依赖一系列对话行为，例如澄清（“你是什么意思？”）和确认（“我明白了。”）。但是，尚不清楚大型语言模型（LLM）生成的文本是否反映了人类的接地行为。为此，我们整理了一组接地行为，并提出了相应的指标，以量化尝试进行的接地。我们研究了LLM生成内容是否包含接地行为，基于多个对话数据集模拟轮流发言，并将结果与人类进行比较。我们发现，与人类相比，LLM生成的语言具有更少的对话接地，而是生成似乎直接假定共同基础已经存在的文本。为了理解所识别出的接地缺口的根源，我们考察了指令微调和偏好优化的作用，发现基于当代偏好数据进行训练会导致生成的接地行为减少。总体而言，我们强调需要开展更多研究，以考察人机交互中的对话接地问题。
英文摘要：
Effective conversation requires common ground: a shared understanding between the participants. Common ground, however, does not emerge spontaneously in conversation. Speakers and listeners work together to both identify and construct a shared basis while avoiding misunderstanding. To accomplish grounding, humans rely on a range of dialogue acts, like clarification (What do you mean?) and acknowledgment (I understand.). However, it is unclear whether large language models (LLMs) generate text that reflects human grounding. To this end, we curate a set of grounding acts and propose corresponding metrics that quantify attempted grounding. We study whether LLM generations contain grounding acts, simulating turn-taking from several dialogue datasets and comparing results to humans. We find that -- compared to humans -- LLMs generate language with less conversational grounding, instead generating text that appears to simply presume common ground. To understand the roots of the identified grounding gap, we examine the role of instruction tuning and preference optimization, finding that training on contemporary preference data leads to a reduction in generated grounding acts. Altogether, we highlight the need for more research investigating conversational grounding in human-AI interaction.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
