大模型
arxiv.org ·
对抗人格漂移的提示策略：比较 LLM 模拟对话中的干预时机与内容
该 arXiv 预印本研究 LLM 模拟学生人格在长对话中的行为漂移问题，采用监控与干预分离的流程，在 1200 组 28 轮对话、4 个 LLM 和两种 ADHD 人格强度下比较五种提示级机制，并区分干预时机（静态 vs 自适应）与干预内容（重新注入 vs 反思提醒），另设由监控器生成行为特定指令的自适应条件。结果显示，相对无干预，重新注入使 LLM 评定的漂移率降低 35–38%，反思提醒降低 22–27%，行为特定指令降低 87%，但均未消除漂移；未发现自适应时机优于静态调度，提示监控更适用于决定“纠正什么”而非“何时干预”。
AI推荐理由 arXiv 预印本，方法描述具体：在 1200 组 28 轮对话、4 个 LLM、两种 ADHD 人格强度下比较五种提示级干预机制，报告 reinjection 降低 LLM 评定的漂移率 35–38%、反思提醒 22–27%、行为特定指令 87%，且均未完全消除漂移，并发现自适应时机未优于静态调度。新信息在于对“干预时机 vs 干预内容”的量化对比，对 LLM 人格模拟与长对话一致性研究有参考价值；局限是漂移率由 LLM 评定、属预印本未经同行评审，结论不能外推为已解决人格漂移问题。
72
正文
Computer Science > Human-Computer Interaction
arXiv:2609.24532 (cs)
[Submitted on 21 Sep 2026]
Title:Prompting Against Persona Drift: Comparing Intervention Timing and Content in LLM-Simulated Conversations
Authors:Nicolas Leins, Jennifer Haase, Varvara Geronimus, Jana Gonnermann-Müller, Sebastian Pokutta
View a PDF of the paper titled Prompting Against Persona Drift: Comparing Intervention Timing and Content in LLM-Simulated Conversations, by Nicolas Leins and 3 other authors
View PDF
HTML (experimental)
Abstract:Simulating student personas with large language models (LLMs) enables scalable evaluation of educational systems. However, behavioral drift, a progressive decline in persona consistency, can emerge over extended conversations, limiting the validity of such simulations. We evaluate five prompt-level mechanisms using separate monitoring and intervention pipelines. Across 1,200 28-turn conversations spanning four LLMs and two ADHD persona intensities, we varied when to intervene (static vs. adaptive) and what to inject (reinjection vs. reflective reminder), plus a novel adaptive condition in which a monitor generates behavior-specific instructions. Relative to no intervention, reinjection reduced the modeled rate of LLM-rated drift by 35--38\%, reflective reminders by 22--27\%, and behavior-specific instruction by 87\%. None eliminated drift. We found no evidence that adaptive timing outperformed static scheduling. Monitoring therefore appears more useful for deciding \textit{what} to correct than \textit{when} to intervene, although behavior-specific instruction requires component-level testing.
Subjects:
Human-Computer Interaction (cs.HC)
Cite as:
arXiv:2609.24532 [cs.HC]
(or
arXiv:2609.24532v1 [cs.HC] for this version)
https://doi.org/10.48550/arXiv.2609.24532
Focus to learn more
arXiv-issued DOI via DataCite (pending registration)
Submission history
From: Nicolas Leins [view email]
[v1]
Mon, 21 Sep 2026 13:05:10 UTC (289 KB)
Full-text links:
Access Paper:
View a PDF of the paper titled Prompting Against Persona Drift: Comparing Intervention Timing and Content in LLM-Simulated Conversations, by Nicolas Leins and 3 other authors
View PDF
HTML (experimental)
TeX Source
view license
Current browse context:
cs.HC
new
|
recent
| 2026-09
Change to browse by:
cs
References & Citations
NASA ADS
Google Scholar
Semantic Scholar
export BibTeX citation
Loading...
BibTeX formatted citation
×
loading...
Data provided by:
Bookmark
Bibliographic Tools
Bibliographic and Citation Tools
Bibliographic Explorer Toggle
Bibliographic Explorer (What is the Explorer?)
Connected Papers Toggle
Connected Papers (What is Connected Papers?)
Litmaps Toggle
Litmaps (What is Litmaps?)
scite.ai Toggle
scite Smart Citations (What are Smart Citations?)
Code, Data, Media
Code, Data and Media Associated with this Article
alphaXiv Toggle
alphaXiv (What is alphaXiv?)
Links to Code Toggle
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub Toggle
DagsHub (What is DagsHub?)
GotitPub Toggle
Gotit.pub (What is GotitPub?)
Huggingface Toggle
Hugging Face (What is Huggingface?)
ScienceCast Toggle
ScienceCast (What is ScienceCast?)
Demos
Demos
Replicate Toggle
Replicate (What is Replicate?)
Spaces Toggle
Hugging Face Spaces (What is Spaces?)
Spaces Toggle
TXYZ.AI (What is TXYZ.AI?)
Related Papers
Recommenders and Search Tools
Link to Influence Flower
Influence Flower (What are Influence Flowers?)
Core recommender toggle
CORE Recommender (What is CORE?)
Author
Venue
Institution
Topic
About arXivLabs
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Which authors of this paper are endorsers? |
Disable MathJax (What is MathJax?)