What it means
Explainability is the degree to which people can understand why an AI system produced a particular output or decision, in terms that are meaningful to them. Explanations can range from showing which factors influenced a result to citing the sources an answer was based on.
Real-world example
A housing office uses AI to help prioritise repair requests. When a tenant's request is marked as low priority, the caseworker can see the main factors (no safety risk reported, similar repair completed recently) and the tenant receives a plain-language reason with a way to provide more information.
Why it matters for AI adoption
People are more likely to trust, use and correctly challenge AI when they can understand its outputs. For large language models, full explanations of internal reasoning are not currently possible, so practical explainability often focuses on sources, confidence, limitations and the factors used in a decision.
Often confused with
Transparency is about openness regarding the system as a whole: that AI is used, what it does, its purpose and its limits. Explainability is about understanding a specific output or decision. A system can be transparent about its use while still giving poor explanations of individual results.
Questions to ask
- What explanation can the system give for a specific output, and to whom?
- Is the explanation useful for the person who needs to act on or challenge the result?