Skip to main content
QUICK REVIEW

[论文解读] PersonaCite: VoC-Grounded Interviewable Agentic Synthetic AI Personas for Verifiable User and Design Research

Mario Truss|arXiv (Cornell University)|Jan 29, 2026
Persona Design and ApplicationsComputer Science被引用 0
一句话总结

PersonaCite 是一个代理系统,在真实 VoC 数据中定位 AI 人物形象,在对话过程中检索证据,并提供带来源、具规避承诺意识的响应以用于可核验的设计研究。

ABSTRACT

LLM-based and agent-based synthetic personas are increasingly used in design and product decision-making, yet prior work shows that prompt-based personas often produce persuasive but unverifiable responses that obscure their evidentiary basis. We present PersonaCite, an agentic system that reframes AI personas as evidence-bounded research instruments through retrieval-augmented interaction. Unlike prior approaches that rely on prompt-based roleplaying, PersonaCite retrieves actual voice-of-customer artifacts during each conversation turn, constrains responses to retrieved evidence, explicitly abstains when evidence is missing, and provides response-level source attribution. Through semi-structured interviews and deployment study with 14 industry experts, we identify preliminary findings on perceived benefits, validity concerns, and design tensions, and propose Persona Provenance Cards as a documentation pattern for responsible AI persona use in human-centered design workflows.

研究动机与目标

  • 通过以经验性 VoC 数据来定位人设,解决基于提示的 AI 人设的不可靠性与不可验证性。
  • 在对话过程中实现对 VoC 工件的实时检索,以约束并验证响应。
  • 引入在证据不足时的显式放弃机制以及对响应层面的来源归属。
  • 通过与行业专家的形成性评估来演示该方法,并概述负责任 AI 人设使用的文档模式。

提出的方法

  • 获取 VoC 工件的检索增强交互,针对每个对话轮次从多模态数据源提取证据。
  • 约束机制仅允许受检索证据支持的 LLM 响应。
  • 当证据缺失时显式放弃,避免推测性主张。
  • 响应层面的来源归属,将主张与底层 VoC 工件关联。
  • 代理式情境工程,提供作为情境证据的合适证据并缓解简短性偏见。
  • 两种交互模式:人物采访和带设计刺激物的反应模拟。
Figure 1. Interviewing a data-grounded synthetic persona. Personas respond only when supported by evidence and explicitly acknowledge gaps in available data. The examples shows a discussion between a designer and the synthetic person. It’s an illustrative brand and was chosen, because it’s widely kn
Figure 1. Interviewing a data-grounded synthetic persona. Personas respond only when supported by evidence and explicitly acknowledge gaps in available data. The examples shows a discussion between a designer and the synthetic person. It’s an illustrative brand and was chosen, because it’s widely kn

实验结果

研究问题

  • RQ1如何将 AI 人设 grounding 在真实 VoC 数据中,以在互动设计研究场景中提升可验证性与信任度?
  • RQ2哪些设计机制(放弃、来源归属、来源证明文档)能在实践中提升 AI 人设的透明度与可靠性?
  • RQ3相比传统方法,检索增强的 grounding 是否支持更快且更负责任的设计探索?
  • RQ4哪些文档模式(Persona Provenance Cards)能帮助在以人为本的工作流程中负责任地部署 AI 人设?

主要发现

  • 通过实时证据检索与响应约束进行 grounding,将人设的定位从创建时转向交互时点。
  • 放弃与来源归属提升了感知的可信度,并支持对人设主张的验证。
  • 反应模拟实现了快速的设计探索与假设检验,同时可与直接用户参与互补。
  • 参与者重视数据来源的透明度,期望有信心分数与将主张与 VoC 数据引用相连的引用。
  • grounding 人设被视为探索性工具,作为对传统用户研究的互补,而非替代。
Figure 2. PersonaCite architecture. VoC data from social media and other channels is imported and stored. Multimodal AI identifies topics and derives personas, storing personas alongside vectorized post data. Persona interaction and reaction simulation with response-level source attribution is enabl
Figure 2. PersonaCite architecture. VoC data from social media and other channels is imported and stored. Multimodal AI identifies topics and derives personas, storing personas alongside vectorized post data. Persona interaction and reaction simulation with response-level source attribution is enabl

阅读本论文

在 Nubint 阅读器中打开原文,边读边就任意段落提问。

在 Nubint 中阅读本论文

本解读由 AI 根据论文原文生成,引用前请核对原文。