大模型会模仿用户说话风格吗？研究发现它们确实会，且常过度模仿。
Do language models accommodate their users? A study of linguistic convergence
- 通过对比模型回复与人类原话，测试语言模型是否随用户风格调整
- 16个模型在3个语料库上均显示强风格趋同，部分超过人类基准
- 指令微调和大模型收敛更弱，说明人机适应机制不同
尽管大型语言模型（LLMs）通常被认为具备出色的语言生成能力，但其语言使用与人类的相似程度仍缺乏深入研究。本文系统评估了模型是否存在语言趋同现象——即模型是否会适应并趋近于用户的语言模式。我们基于16个语言模型、3个对话语料库及多种风格特征，比较了模型生成回复与原始人类回应的一致性。结果表明，模型在对话中表现出强烈的风格趋同，其趋同程度往往显著高于人类基线。虽然趋同模式因特征而异，但在不同建模设置下均观察到一致的趋同趋势：经过指令微调和参数更大的模型，其趋同程度低于预训练和较小的模型。鉴于人类与模型在趋同模式上的差异，我们推测其背后的驱动机制存在本质不同。
原文摘要 · Abstract (English)
While large language models (LLMs) are generally considered proficient in generating language, how similar their language usage is to that of humans remains understudied. In this paper, we test whether models exhibit linguistic convergence, a core pragmatic element of human language communication: do models adapt, or converge, to the linguistic patterns of their user? To answer this, we systematically compare model completions of existing dialogues to original human responses across sixteen language models, three dialogue corpora, and various stylometric features. We find that models strongly converge to the conversation's style, often significantly overfitting relative to the human baseline. While convergence patterns are often feature-specific, we observe consistent shifts in convergence across modeling settings, with instruction-tuned and larger models converging less than their pretrained and smaller counterparts. Given the differences in human and model convergence patterns, we hypothesize that the underlying mechanisms driving these behaviors are very different.
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。