Symbolic AI
AI built on explicit symbols and logical rules — the original vision of reasoning through manipulation of formal, human-readable representations.
Think of it like
Like doing math on paper: each symbol is explicit, each rule is stated outright, and you can follow the logic step by step.
Example
Early AI researchers thought they could encode human knowledge as formal logic: "All men are mortal. Socrates is a man. Therefore, Socrates is mortal." They built systems that chained such logical statements to derive new conclusions.
How it actually works
The symbolic AI worldview assumes intelligence is symbol manipulation — that thinking is reasoning with explicit, discrete symbols bound by formal rules. This contrasted with connectionism (neural networks), which emerged later and said reasoning emerges from sub-symbolic, learned patterns in numbers. Symbolic AI dominated research through the 1980s but hit limits: it couldn't learn from data efficiently, struggled with nuance and exception handling, and required brittle manual knowledge engineering. Modern LLMs are sometimes called "sub-symbolic" because they work with learned numerical patterns, not explicit rules.
For product teams
A worthy first attempt, but the requirement to hand-code all domain knowledge made it expensive and brittle.
For engineers
An approach to AI based on formal logic, explicit rules, and symbolic manipulation, typically implemented as rule engines or theorem provers.
Related
- Neural network — Foundational approach before connectionism and neural networks.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome