Caesar AI Atlas
MétricasIntermedio

Unsupported-claim Rate vs Hallucination

A side-by-side comparison of Unsupported-claim Rate and Hallucination. Understand how a measurement of unsupported claims relates to the broader failure of false or fabricated AI output.

Veredicto rápido: Use Unsupported-claim Rate when quantifying unsupported assertions; use Hallucination when describing the underlying generative AI failure mode.

De un vistazo

Unsupported-claim Rate

Unsupported-claim Rate describes percentage of claims in a model response that are not grounded in supporting evidence.

Características clave
  • Percentage-based evaluation metric
  • Focuses on claims not grounded in evidence
  • Useful for response-level quality measurement
  • Signals risk of unsupported assertions
Tener en cuenta
  • Counting claims inconsistently
  • Treating the metric as proof of factual accuracy
  • Ignoring evidence quality when assessing support

Contexto: Most relevant for evaluation reports that need a measurable rate of unsupported claims.

VS
Hallucination

Hallucination describes AI-generated output that appears plausible or confident but is false, unsupported, misleading, or fabricated.

Características clave
  • False, unsupported, misleading, or fabricated output
  • Often appears plausible or confident
  • May include invented facts or sources
  • Mitigated through grounding and verification
Tener en cuenta
  • Using the label for every model error
  • Ignoring confident but unsupported citations
  • Assuming grounding eliminates hallucinations completely

Contexto: Most relevant when describing a generative AI reliability or safety failure.

Diferencias clave

AspectoUnsupported-claim RateHallucination
What it measuresUnsupported-claim rate measures the percentage of claims in a response that lack supporting evidence.Hallucination describes a type of output that is false, unsupported, misleading, or fabricated.
Best use caseUse UCR to compare systems, prompts, or retrieval settings using a defined evaluation rubric.Use hallucination to describe the failure mode observed in a model response or incident.
Failure modeA high UCR indicates that many response claims are not grounded in evidence.A hallucination may be one unsupported claim, an invented citation, or a broader fabricated explanation.
Threshold sensitivityUCR depends on how claims are segmented, what counts as evidence, and which threshold is used for support.Hallucination assessment depends on factual verification, grounding checks, and the task context.
Common mistakeA common mistake is treating a low UCR as proof that every answer is correct.A common mistake is using hallucination as a vague label without recording which claims were false or unsupported.
Nota Caesar AI

In practice, hallucination is the problem category, while unsupported-claim rate is one way to measure and govern part of that problem.

Notas

Errores comunes

1

Reporting hallucination risk without a measurable evaluation method.

2

Counting unsupported claims without defining what qualifies as a claim or supporting evidence.

3

Assuming citations are valid without checking whether they support the statement.

4

Using aggregate UCR to hide severe individual failures.

Cuándo usar cada uno

unsupported-claim-rate

Use Unsupported-claim Rate when evaluating responses against source material or a grounding standard. It is appropriate for dashboards, regression tests, and evidence packs where unsupported assertions need to be counted.

hallucination

Use Hallucination when describing the substantive error pattern in generative AI output. It is the right term for user-facing risks, incident descriptions, and reliability controls.

Nota de cumplimiento

For ISO/IEC 42001 and NIST AI RMF evidence, unsupported-claim metrics can support monitoring and validation of hallucination risk, especially where AI outputs inform decisions or external communications.

Preguntas frecuentes

Is unsupported-claim rate the same as hallucination rate?+

No. Unsupported-claim rate measures unsupported claims in a response, while hallucination is a broader failure category that may include false facts, fabricated sources, and misleading reasoning.

Can a supported claim still be wrong?+

Yes. A claim may cite a source but still misread it or rely on unreliable evidence. Grounding and factuality should be evaluated separately where the use case requires it.

Why is UCR useful for governance?+

It turns an abstract hallucination risk into a repeatable measurement. That makes it easier to track changes across prompts, models, retrieval settings, and releases.

Vistos recientemente

No recently viewed comparisons yet.