arXiv：cs.CL· Indu Panigrahi, Tal August·· 16 小时前
让 LLM 生成可交互调节复杂度的科学摘要：五项模型最多 33.67% 响应方向正确
Explain Like I'm 5 or Whatever I Choose: Evaluating the Potential for Interactive Complexity in Language Model Responses
AI 导读
一项用户研究评估了五个不同家族、规模和推理能力的 LLM 在 557 条科学查询上按复杂度连续生成摘要的能力，发现模型虽能改变文本复杂度，但大多数变化不一致，复杂度指标朝正确方向移动的比例最高仅 33.67%。研究通过被试内实验确定了可靠的复杂度度量指标及模型应遵循的判定标准，目标是为用户提供直接操纵文本复杂度的控制机制。
正文
Abstract:Given the increased use of large language models (LLMs) for reading and seeking information and the varying needs and backgrounds of users, direct manipulation of text can be a valuable mechanism for end-user control. However, enabling direct manipulation requires that models generate text consistent with user-requested changes. We focus on one such setting, manipulating the complexity of scientific information, and ask if models can generate multiple summaries along a continuum of text complexity that enable effective user selection and control. Through a within-subjects user study, we identify reliable complexity measures and a criterion for how models should conform to these measures. We evaluate five LLMs of different families, sizes, and reasoning abilities on $557$ scientific queries. We find that while models can vary complexity across responses, most changes remain inconsistent, with models shifting the complexity measures in the correct direction at most $33.67$% of the time.
| Comments: | Preprint |
| Subjects: | Computation and Language (cs.CL); Human-Computer Interaction (cs.HC) |
| Cite as: | arXiv:2606.06788 [cs.CL] |
| (or arXiv:2606.06788v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2606.06788 arXiv-issued DOI via DataCite |
Submission history
From: Indu Panigrahi [view email]
[v1]
Fri, 5 Jun 2026 00:14:44 UTC (4,050 KB)
[v2]
Wed, 14 Oct 2026 13:55:08 UTC (4,071 KB)
来源：arXiv：cs.CL · arxiv.org