世界经济论坛＆amp凯捷：2025已付实践的AI智能体：评估与治理基础白皮书（英文版）（34页）.pdf

1、AI Agents in Action:Foundations for Evaluation and GovernanceW H I T E P A P E RN O V E M B E R 2 0 2 5In collaboration with Capgemini Images:Adobe StockDisclaimer This document is published by the World Economic Forum as a contribution to a project,insight area or interaction.The findings,interpret

2、ations and conclusions expressed herein are a result of a collaborative process facilitated and endorsed by the World Economic Forum but whose results do not necessarily represent the views of the World Economic Forum,nor the entirety of its Members,Partners or other stakeholders.2025 World Economic

3、 Forum.All rights reserved.No part of this publication may be reproduced or transmitted in any form or by any means,including photocopying and recording,or by any information storage and retrieval system.ContentsForeword 4Executive summary 5Introduction 61 Evolving technical foundations of AI agents

4、 81.1 The software architecture of an AI agent 81.2 Communication protocols and interoperability 101.3 Cybersecurity considerations 122 Foundations for AI agent evaluation and governance 132.1 Classification 142.2 Evaluation 192.3 Risk assessment 222.4 Governance considerations for AI agents:a progr

5、essive approach 253 Looking ahead:multi-agent ecosystems 29Conclusion 30Contributors 31Endnotes 34AI Agents in Action:Foundations for Evaluation and Governance2ForewordIn recent years,organizations have moved beyond predictive models and chat interfaces to experiment with artificial intelligence(AI)

6、in more transformative ways.AI agents are now emerging as integrated collaborators in business,public services and everyday life.The adoption of AI agents could bring significant gains in efficiency,altered kinds of human-machine interaction and the advent of novel digital ecosystems.This transition