Ontologies in the Age of AI: Why Better Models Still Depend on Better Meaning

2026-01-05 · Source: Modern Data 101 · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Intermediate, medium

Summary

Sonam Bijani, a Data Quality & Governance leader, argues that improving AI outcomes in enterprises, particularly within investment data programs, hinges on establishing strong data foundations rather than solely focusing on advanced model capabilities. She highlights "semantic fragility," where terms like "portfolio" carry multiple, valid meanings across different internal systems (e.g., fund accounting, order management, performance, risk systems). Large language models (LLMs) operating on statistical patterns struggle with this ambiguity, leading to "semantic blending" rather than true understanding, which can produce numerically plausible but conceptually incorrect outputs. Traditional data architectures, including data warehouses and master data management, are often insufficient because they describe meaning rather than formally encode it in a machine-interpretable structure. Bijani advocates for an ontology layer to explicitly model classes, relationships, hierarchies, and constraints, thereby narrowing the LLM's hypothesis space and grounding its reasoning in defined meaning.

Key takeaway

For CTOs and VPs of Engineering/Data evaluating enterprise AI deployments, prioritize investing in formal conceptual modeling and ontology layers over solely pursuing advanced LLM capabilities. Your organization's ability to explicitly define and reconcile core business entities like "portfolio" will directly determine the reliability and trustworthiness of AI-generated insights, mitigating the risk of subtle but dangerous "semantic blending" in critical financial contexts.

Key insights

Semantic clarity, formalized through ontologies, is critical for reliable enterprise AI outcomes, especially in complex domains like finance.

Principles

• AI outcomes depend on data foundations.
• Semantic ambiguity scales with AI fluency.
• Ontologies encode meaning for AI systems.

Method

Implement an ontology layer to formally represent classes, relationships, hierarchies, and constraints, mediating between structured data and LLM interfaces to ensure defined meaning over probabilistic inference.

In practice

• Define core business concepts formally.
• Model entity relationships explicitly.
• Integrate ontologies into AI architecture.

Topics

• Ontologies
• Semantic Clarity
• Large Language Models
• Enterprise AI
• Data Governance

Best for: CTO, VP of Engineering/Data, Executive, AI Architect, Data Engineer, Director of AI/ML

Related on AIssential

• Why machines that can write poetry still can’t agree on what “good” means — Ontology-based semantic layers are crucial for grounding agentic AI in explicit, governed business meaning, overcoming inherent linguistic ambiguity.
• From Models to Momentum: Uniting Architects and Engineers with ER/Studio — Enterprise data modeling provides a semantic backbone, crucial for data reliability, governance, and effective AI integration.
• Why Enterprise AI Needs a Governed Meaning Layer: Introducing Snowflake Horizon Context — Enterprise AI requires a governed meaning layer like Snowflake Horizon Context to resolve semantic ambiguity and prevent hallucinations.
• Semantic Layers in the Age of AI are 100% Needed — Semantic layers translate natural language queries into data operations, ensuring consistent data interpretation for AI and users.
• AI needs a strong data fabric to deliver business value — Enterprise AI success hinges on data context and judgment, not just speed, necessitating a robust data fabric.
• Ontologies Are the Intentional Core of a True Knowledge Graph — Formal ontologies provide the stable, intentional semantic backbone essential for reliable enterprise AI and knowledge graphs.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Modern Data 101.