Responsible AI / Safety · Level 2 of 5
Explainability
The ability to provide useful accounts of a model's behavior or predictions.
An explanation can be persuasive without being faithful or complete.
Example
A report describes which measured inputs influenced a prediction.
Listen to the definition and example
Audio transcript
Explainability. The ability to provide useful accounts of a model's behavior or predictions. An explanation can be persuasive without being faithful or complete. For example: A report describes which measured inputs influenced a prediction.
Explore this concept
Why it matters
This helps you identify a concrete failure mode or evaluate the limits of a proposed control.
Start with
Related concepts
Quick recall question
Try answering before looking back at the definition.