---
id: "concept-semantic-vs-functional-correctness"
type: "concept"
source_timestamps: ["00:18:18", "00:18:51"]
tags: ["quality-assurance", "evaluation"]
related: ["concept-guardrails-security-design", "concept-silent-failure", "concept-evaluation-quality-judgment"]
definition: "The distinction between an AI output that merely sounds plausible (semantic) versus one that is factually accurate and successfully executable (functional)."
sources: ["s42-job-market-split"]
sourceVaultSlug: "s42-job-market-split"
originDay: 42
---
# Semantic vs. Functional Correctness

## Critical distinction

A critical distinction in verifying AI outputs:

- **Semantic correctness**: the LLM's output sounds right and is logically coherent in text. *'Here is the right credit card for you.'*
- **Functional correctness**: the output is actually, factually true and executable in the real world. The recommended credit card is, in fact, the correct product for that specific user's data.

## Why it matters

Tolerating only semantic correctness is exactly how [[concept-silent-failure-d42]] is born. Production systems must be measured against **functional correctness** — see [[concept-evaluation-quality-judgment]] and [[action-build-eval-harnesses]].

This distinction is one of the verifiability inputs to [[concept-guardrails-security-design]].