Enhancing LLMs with Graph Technology
summary
Claims that combining LLMs with graph databases reduces hallucinations and adds explainability, supported by demos using Neo4j, LangChain, OpenAI models and a visual Knowledge Graph Builder that ingests SPDX docs and YouTube transcripts. Shows math-and-bias failures from ChatGPT 3.5, GPT‑4/4o and a slow 0.3 model where humans beat LLMs on a seating grid, cites Gartner numbers (30% of gen AI projects abandoned after POC; 50% improved reliability via graphs) and production examples like Clara and Pfizer.
AI-written summary · about
chapters (8)
- 0:00Introduction and overview
- 3:10Limitations of LLMs
- 19:30Why hallucinations occur
- 23:19Knowledge graphs explained
- 29:20Combining graphs with LLMs
- 32:09Graph RAG architecture
- 36:23Demo and tools
- 40:12Conclusion and resources
related talks
44mGive Your LLMs a Left Brain
28mEnhancing LLMs with Graph Technology
52mGive Your LLMs a Left Brain
28mGrounded Intelligence Boosting LLMs with Real-Time Graph Context
23mKnowledge Graphs and LLMs in Action
28mThe Present and Future of AI-Enabled Knowledge Graph Generation
27mLies, Damn Lies, and AIs
47mGive Your LLMs a Left Brain