威胁情报中最小人工智能原则：有效采用全人类人工智能的蓝图.pdf

1、Principle of Least AI in Threat Intelligence:A Blueprint for Effective GenAI AdoptionJitin ShabaduForrester Security Analyst2 Forrester Research,Inc.All rights reserved.Vendor Approaches to building AI AgentsRick HollandBrian KimeBrian WrozekJitin ShabaduJosh Zelonis3 Forrester Research,Inc.All righ

2、ts reserved.Vendor Approaches to building AI AgentsHe is an analyst supporting security and risk professionals.His research focuses on Zero Trust,network analysis and visibility(NAV),threat intelligence,deception technology,and automation and orchestration in Zero Trust environments.He works with cl

3、ients to enhance their organizations security posture through Zero Trust and counter cyberthreats through automation and orchestration strategies.Jitins expertise spans numerous industries,including energy,finance,aviation,education,and government,enabling him to deliver tailored solutions for compl

4、ex security challenges.Jitin ShabaduEducationWork Experience Forrester Research,Inc.All rights reserved.Current Market For AI Adoption4 Forrester Research,Inc.All rights reserved.Plain-English AI:The Definitions That CountUse Of Backend Machine Learning(ML)Wrapping Of LLMs“AI Agent”via Prompt Sequen

5、cesUsable AI Agent-ish CapabilitiesAgentic AI Forrester Research,Inc.All rights reserved.ORVendor Approaches to building AI AgentsHuman Role-specific AI AgentTask-specific AI Agent Forrester Research,Inc.All rights reserved.Vendor Approaches to building AI AgentsPros:1.Better suited as orchestrators

6、 for other AI Agents2.Enhances workflows(e.g.,minimizes tool switching)Cons:1.Relatively lower on utility and often lacks clarity in scope of functionality and ownership2.Does not fully replicate the intended role in its entirety and can be hard to measure and evaluateHuman Role-specific AI Agent Fo