The White House Office of Science and Technology Policy has outlined five principles that should be used as guidance in designing, deploying and using artificial intelligence and other automated systems to protect the civil rights of U.S. citizens and democratic values.
OSTP said the principles under the Blueprint for an AI Bill of Rights are safe and effective systems; algorithmic discrimination protections; data privacy; notice and explanation; and human alternatives, consideration and fallback.
For the first principle, automated systems should go to pre-deployment testing, risk identification and mitigation and continuous monitoring to ensure that such platforms are safe and effective.
To prevent discrimination by algorithms and related systems, the document calls for designers and developers to take measures such as performing proactive equity evaluation as part of the system design, ensuring accessibility for people with disabilities during the design, development and pre-deployment phases, using representative data and protecting against proxies for demographic features.
