A team places importance on having documented information about an AI model it uses, such as what data it was trained on, its intended uses, and its known limitations. Which responsible AI characteristic BEST matches this property?
A question about choosing the responsible AI characteristic where the model's profile is disclosed.
Explainability
Explainability is the property of being able to understand why an individual prediction was reached.
This question is about disclosing the model's overall profile (data, uses, limitations), not the reasoning of an individual prediction, so this is incorrect.
Robustness
Robustness is the property of operating stably against input variation and attacks.
This question is about disclosing the model's profile, not strength against input variation, so this is incorrect.
Controllability
Controllability is the property of humans being able to oversee the AI's behavior and intervene or stop it as needed.
This question is about disclosing the model's profile, not controlling its behavior, so this is incorrect.
Transparency
Correct. Transparency is the property of having the model's training data, intended uses, known limitations, and similar information disclosed. Making the model's profile clear with documents such as SageMaker Model Cards corresponds to this.
Disclosing the training data, uses, and limitations of the model as a whole is transparency, and SageMaker Model Cards is a representative way to achieve it. It is easy to confuse with explainability, which is about understanding the reasoning of an individual prediction, so remember: overall disclosure = transparency / individual reasoning = explainability.