The White House recently released its Blueprint for an AI Bill of Rights in an effort to guide the discussion on the design, use and deployment of AI in systems that impact the American public. The Blueprint outlines the following five guiding principles:
- Safe and Effective Systems: Systems should be developed and monitored to ensure they are safe and effective. This should include independent evaluation and reporting.
- Algorithmic Discrimination Protections: Systems should be continuously monitored to protect against algorithmic discrimination. This may include equity assessments, disparity testing, and organizational oversight.
- Data Privacy: Privacy by design should guide the systems. Privacy protections should seek individual permission and respect choices made regarding the collection, use, access, transfer and deletion of data.
- Notice and Explanation: Use of AI should be explained in a manner that is clear, timely, and accessible.
- Human Alternatives, Considerations, and Fallback: Individuals should have the option to opt out of AI systems in favor of a human alternative.
This Blueprint is non-binding on companies and provides only a starting point on the discussion of how best to protect the American public from potential harms from the use of automated systems. A technical companion included with the Blueprint provides examples and steps companies can use to implement the principles.
Putting it Into Practice: While not binding, this blueprint is a reminder that AI is receiving increasing scrutiny, which we anticipate will continue in the United States and worldwide.