Glossary · Technical
Confidence Calibration
How well a model's expressed certainty matches the actual likelihood that its answer is correct.
What it means
A well-calibrated model says "I'm not sure" roughly as often as it's actually wrong, and states things confidently roughly as often as it's actually right. A poorly calibrated model can sound equally confident whether it's citing a verified fact or repeating something it hallucinated.
Why it matters for AI visibility
Fluent, assertive phrasing isn't a signal of accuracy — a model can state an outdated price, a discontinued feature, or a fabricated claim about a brand with the exact same confident tone it uses for something verified. That's part of why AI visibility monitoring has to check not just whether a brand is mentioned, but whether what's said about it is actually true.
See more terms in the full glossary, or read the AI Visibility 101 guide for the full picture.