---
title: 'Expert system'
description: "A bounded AI system that applies an explicit knowledge base and inference procedure to reproduce part of a specialist's decision process."
canonical_url: 'https://darkfactory.dev/glossary/expert-system'
markdown_url: 'https://darkfactory.dev/glossary/expert-system.md'
collection: glossary
date_published: '2026-08-05T00:00:00-04:00'
date_modified: '2026-08-26T00:00:00-04:00'
---

# Expert system


## Definition

An expert system is a bounded AI system that applies an explicit knowledge base and inference procedure to reproduce part of a specialist's decision process. Classic systems separate domain facts and heuristics from an inference engine that evaluates rules, maintains intermediate conclusions, and may explain which rule path produced a recommendation.

The knowledge base can contain facts, if-then rules, frames, certainty factors, or other symbolic representations. The system's competence comes from the captured domain knowledge and from how the inference engine applies it. That makes the boundary easier to inspect than a learned neural model, but the knowledge still has to be elicited, encoded, reviewed, and kept current.

## Historical example

DENDRAL, developed at Stanford beginning in the 1960s by Edward Feigenbaum, Joshua Lederberg, Bruce Buchanan, Georgia Sutherland, and collaborators, is widely documented as the first expert system. It used chemical knowledge and heuristic search to propose molecular structures consistent with mass-spectrometry data.

The historical record supports DENDRAL's place in the field. It does not establish one person as the proven coiner of the phrase "expert system."

## Distinguish it from nearby terms

- A **rule engine** executes explicit rules but may make no claim to encode specialist knowledge or solve an expert task.
- A **machine-learning model** learns parameters from data. An expert system traditionally receives much of its knowledge through explicit representation.
- A **decision-support system** may organize evidence for a person without applying an expert knowledge base.
- A **hybrid AI system** can use learned models for perception or prediction and explicit rules for policy, constraints, or explanation.

## Operational significance

An explanation trace proves which rules fired, not that the conclusion was correct. Expert systems fail when knowledge is missing, stale, contradictory, poorly scoped, or applied to a case outside the domain. Governance therefore centers on knowledge ownership, versioning, conflict resolution, and an explicit route for cases the rules do not cover.

## Check your understanding

A diagnostic rule says a finding implies one disease, but the medical guideline changed last month. The inference trace may be perfectly clear while the recommendation is wrong because the knowledge base is stale.

## Related terms

- [AI system](https://darkfactory.dev/glossary/ai-system)
- [Narrow AI](https://darkfactory.dev/glossary/narrow-ai)
- [Knowledge graph](https://darkfactory.dev/glossary/knowledge-graph)
- [Explainability](https://darkfactory.dev/glossary/explainability)

## Evidence and further reading

- [Andreessen Horowitz AI Glossary](https://a16z.com/ai-glossary/)
- [Stanford HAI Artificial Intelligence Glossary](https://hai.stanford.edu/ai-definitions)
- [DENDRAL Software Preservation Project](https://softwarepreservation.computerhistory.org/AI/DENDRAL/)