Extrinsic Trust as a Contractual Framework for Accountable AI in Health Care: Viewpoint.

Area of Science:

• Healthcare technology
• Artificial intelligence in medicine
• Clinical informatics

Background:

• Artificial intelligence (AI) offers potential for healthcare efficiency and equity.
• Widespread AI adoption in healthcare is hindered by insufficient trust.
• A gap exists between intrinsic trust (interpretability) and extrinsic trust (functional validation).

Purpose of the Study:

• To address the trust deficit in healthcare AI adoption.
• To propose a novel contractual framework for trustworthy AI.
• To guide the operationalization of AI trust in clinical settings.

Main Methods:

• A viewpoint analysis of trust in AI within healthcare.
• Development of a contractual framework with three core promises: reliability, scope and equity, and shift and uncertainty.
• Illustration of the framework's application through a clinical vignette.

Main Results:

• The proposed framework bridges the gap between intrinsic and extrinsic trust.
• The framework operationalizes trust through structured evidence and governance.
• The vignette demonstrates translating trustworthy AI principles into accountable clinical deployment.

Conclusions:

• A contractual framework is essential for fostering trust in healthcare AI.
• Implementing promises of reliability, scope, equity, and uncertainty management is key.
• This approach facilitates the accountable clinical deployment of trustworthy AI systems.