用人格模型控制大模型代理，让其行为更真实可信。
Personality-Driven Decision-Making in LLM-Based Autonomous Agents
- 基于OCEAN人格模型诱导代理性格，影响任务选择决策
- 不同人格特征对应明显不同的任务规划与调度模式
- 适用于主动网络安全防御等需拟人化行为的场景
将大语言模型（LLMs）嵌入自主代理是快速发展的领域，可实现无需特定领域训练的动态、可配置行为。在前期工作中，我们提出了SANDMAN——一种利用五因素人格模型（OCEAN）的欺骗性代理架构，证明了人格诱导显著影响代理的任务规划。本研究进一步提出一种新方法，用于衡量和评估诱导人格特质如何影响基于LLM的代理在任务选择过程中的表现，具体涵盖规划、调度与决策。实验结果揭示了与诱导的OCEAN属性一致的任务选择模式，证实了设计高度可信的欺骗性代理以支持主动网络防御策略的可行性。
原文摘要 · Abstract (English)
The embedding of Large Language Models (LLMs) into autonomous agents is a rapidly developing field which enables dynamic, configurable behaviours without the need for extensive domain-specific training. In our previous work, we introduced SANDMAN, a Deceptive Agent architecture leveraging the Five-Factor OCEAN personality model, demonstrating that personality induction significantly influences agent task planning. Building on these findings, this study presents a novel method for measuring and evaluating how induced personality traits affect task selection processes - specifically planning, scheduling, and decision-making - in LLM-based agents. Our results reveal distinct task-selection patterns aligned with induced OCEAN attributes, underscoring the feasibility of designing highly plausible Deceptive Agents for proactive cyber defense strategies.
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。