ModelRefs / Uncertainty Quantification — AI Glossary
Uncertainty Quantification — AI Glossary
Methods for estimating how confident a model is in its outputs, distinguishing epistemic (model) from aleatoric (data) uncertainty.
Overview
LLM uncertainty estimation: logprob aggregation (perplexity of the generated sequence), verbalized confidence ('I am 80% confident that…'), sampling-based (how consistent are N samples?), and conformal prediction for calibrated coverage guarantees. Critical for high-stakes applications where knowing when not to trust the model is as important as the answer itself.
Reference details
| Topic | evaluation |
|---|---|
| Also known as | confidence estimation, model confidence |
| Last reviewed | 2026-06-24 |
Related terms
Continue your research
Use these connected ModelRefs sections to compare alternatives, inspect implementation paths, and review the evidence and governance boundaries relevant to Uncertainty Quantification — AI Glossary.
Frequently asked questions
What is Uncertainty Quantification?
Methods for estimating how confident a model is in its outputs, distinguishing epistemic (model) from aleatoric (data) uncertainty.
Is Uncertainty Quantification the same as confidence estimation?
Yes — confidence estimation, model confidence are common aliases for Uncertainty Quantification.
What concepts are related to Uncertainty Quantification?
Closely related concepts include calibration, logprobs, abstention.