面向人工智能和智能系统的知识图谱（由 Neo4j 赞助）.pdf

1、 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.Knowledge Graphs for AI and Intelligent SystemsD A T 2 0 9-S Agentic AI for the EnterpriseMaking A

2、I Trusted,Accurate&ExplainableAgendaThe AI Challenge w/Enterprise Data1Understanding Context Engineering2Why Graphs for AI?3Show by Example-Employee Agent4Neo4j Aura Agent&MCP5Wrap-up&Q&A6The AI Challenge w/Enterprise DataAI is an accelerant for new experiences,efficiencies,and processes Growrevenue

3、ImproveexperiencesMonitorcomplianceReducecomplexityLowercostManageriskDriveinnovationMarketdifferentiationDATA CHALLENGES95%of projects never reach production*Source:Gartner*MIT report:The GenAI Divide Why Enterprise AI is Failing1.AI Hallucinates LLMs make mistakes.In the enterprise,thats often not

4、 good enough.2.AI is a Black Box Very hard to build trust with stakeholders,regulators,and customers if you cant explain anything about AI decisions3.LLMs have no discernment Any data is fair game to AI models.No ability to apply privacy,security,or other restrictions on appropriate,legal,and ethica

5、l data use.4.AI lacks context Models dont have access to relevant enterprise context for each decision85%of IT leaders cite data quality as their top AI challengeKPMG AI Quarterly Pulse Survey 2025100 senior executives from$1B+revenue companiesLack of StandardizationScale of Query DiversityData Orga

6、nizationAI needs data organized in ways it can effectively accessAI requires high-quality harmonized data to be reliableAI demands rapidly accelerated question-to-query executionIts a data management issue!Why Graphs for Trustworthy AI?Boost AccuracyImprove ExplainabilityFuture-Proof AI SystemsGraph