跳转至

文章背景与核心概要

随着自主智能体AI（Agentic AI）和基于大语言模型（LLM）的多智能体系统（MAS）在自动化复杂任务中变得越来越普及，确保人类中心价值观（如隐私、公平和安全性）的保留变得愈发关键。传统的软件工程主要关注功能正确性，而社会-技术AI的深度融合则要求通过深思熟虑的架构选择来实现稳健的价值对齐。

本文探讨了架构模式——具体包括协调机制、通信协议和系统拓扑——如何能够促进多智能体系统中的价值导向行为。作者提出了三种核心的价值保留架构模式：1. 利用联邦拓扑的隐私感知架构；2. 旨在促进多元化和多样性的分布式架构；3. 配置用于检测和缓解不公平现象的守护智能体架构。这些研究为设计值得信赖的多智能体系统奠定了统一的架构模式和指南基础。

Value-Preserving Architectures for Agentic AI Systems

Summary

As agentic AI and Large Language Model (LLM)-based multi-agent systems (MAS) become more prevalent in automating complex tasks, ensuring the preservation of human-centered values (such as privacy, fairness, and safety) is increasingly critical. Traditional software engineering has focused primarily on functional correctness, but socio-technical AI integration demands robust value alignment through deliberate architectural choices.

This paper investigates how architectural patterns—specifically coordination mechanisms, communication protocols, and system topologies—can promote value-oriented behaviors in MAS. The authors propose three core value-preserving architectural patterns: 1. Privacy-aware architecture utilizing a federated topology. 2. Distributed architecture designed to promote pluralism and diversity. 3. Guard-agent architecture configured to detect and mitigate unfairness.

Document Metadata

字段 详情
标题 智能体AI系统的价值保留架构 (Value-Preserving Architectures for Agentic AI Systems)
作者 Alessandro Pesare, Tommaso Dolci, Katja Hose, Emanuel Sallinger
提交时间 2026年9月3日
主学科领域 人工智能 (cs.AI)
会议收录 已被 ASE 2026 AgenticDev Workshop 接受
标识符 arXiv: 2609.03920 [cs.AI]
DOI 10.48550/arXiv.2609.03920
许可协议 知识共享署名 4.0 国际版 license icon
Field Details
Title Value-Preserving Architectures for Agentic AI Systems
Authors Alessandro Pesare, Tommaso Dolci, Katja Hose, Emanuel Sallinger
Submitted September 3, 2026
Primary Subject Artificial Intelligence (cs.AI)
Conference Accepted to the AgenticDev Workshop at ASE 2026
Identifiers arXiv: 2609.03920 [cs.AI]
DOI 10.48550/arXiv.2609.03920
License Creative Commons Attribution 4.0 International license icon

Abstract

智能体AI和基于大语言模型的多智能体系统（MAS）的涌现，为自动化复杂任务带来了前所未有的机遇，同时也引发了人们对保护人类核心价值观（如隐私、公平和安全）的紧迫关切。尽管传统软件工程主要侧重于功能正确性，但在复杂社会-技术系统中采用大语言模型和AI智能体，已经加剧了对负责任的软件工程以及稳健价值对齐的需求。

在多智能体系统中，架构设计决策（如协调机制、通信协议和系统拓扑）在塑造系统行为及其产生的产出方面发挥着核心作用。本文认为，架构选择不仅会影响多智能体系统的功能和性能，还能促进价值导向的系统行为。因此，我们研究了不同的架构设计如何支持各种以人为本的价值观，并探讨了以下价值保留架构模式： * (i) 采用联邦拓扑的隐私感知架构； * (ii) 旨在促进多元化和多样性的分布式架构；以及 * (iii) 用于检测和减轻不公平现象的守护智能体架构。

最后，我们引入了具有代表性的用例，以说明真实场景中提出的架构。通过将架构设计与以人为本的价值观相联系，这项工作为构建可信多智能体系统的统一架构模式和指南奠定了基础。

The emergence of agentic AI and LLM-based multi-agent systems (MAS) presents unprecedented opportunities for automating complex tasks, while simultaneously raising critical concerns about the preservation of fundamental human-centered values, such as privacy, fairness, and safety. Although software engineering has traditionally focused on functional correctness, the adoption of LLMs and AI agents into complex socio-technical systems has intensified the need for responsible software engineering and robust value alignment.

In MAS, architectural design decisions, such as coordination mechanisms, communication protocols, and system topologies, play a central role in shaping system behavior and the outcomes they produce. This paper argues that architectural choices influence not only the functionality and performance of MAS but can also promote value-oriented system behavior. Therefore, we investigate how different architectural designs support different human-centered values, discussing the following value-preserving architectural patterns: * (i) A privacy-aware architecture with a federated topology, * (ii) A distributed architecture to promote pluralism and diversity, and * (iii) A guard-agent architecture to detect and mitigate unfairness.

Finally, we introduce representative use cases to illustrate the proposed architectures in real-world scenarios. By linking architectural design with human-centered values, this work lays the foundation for a unified set of architectural patterns and guidelines towards the design of trustworthy MAS.

Access & Resources