The Joe Reis Show The "Oh Word" - Why Ontologies Are Suddenly Popular
8 snips
Dec 14, 2025 Matt Housley, a seasoned knowledge-engineering and ontology practitioner, dives into the rising buzz around ontologies and knowledge graphs. He discusses how these structures are essential for adding context and semantics to AI systems. The duo highlights the limitations of early language models without knowledge graphs, while tracing the roots of this trend back to Google’s influential knowledge graph. They also caution against the influx of self-proclaimed experts and emphasize the need for foundational knowledge in the field.
AI Snips
Chapters
Transcript
Episode notes
Ontologies Provide Critical Context For Models
- Ontologies are suddenly mainstream because models need structured context and semantics to work well.
- Knowledge graphs and ontologies provide the relationships and memory that improve model behavior.
Early Research Showed LLMs Lacked Context
- Joe recounts a mutual friend's early research showing LLMs failed without knowledge graph context.
- That research at the Alan Turing Institute two years ago demonstrated poor retrieval without structured context.
Knowledge And Engineering Teams Are Converging
- Joe notes knowledge practitioners were separate from engineers but are now converging with data and software teams.
- He gives examples like Mariah Peterson being tasked to implement ontologies at her company.