The National Telecommunications and Information Administration (NTIA) has issued a new request for comment that is looking for input on what policies can support AI accountability and build trust in AI systems.
While AI technologies offer a multitude of benefits, NTIA wants to create earned trust in AI systems and ensure that they do not lead to harmful outcomes.
“Responsible AI systems could bring enormous benefits, but only if we address their potential consequences and harms. For these systems to reach their full potential, companies and consumers need to be able to trust them,” NTIA Administrator Alan Davidson said in an April 11 press release.
“Our inquiry will inform policies to support AI audits, risk and safety assessments, certifications, and other tools that can create earned trust in AI systems,” Davidson added.
Just as financial audits build trust in financial statements, NTIA hopes to do the same to build trust in AI systems. Additionally, just as food and cars go through safety inspections, the agency wants AI to provide assurance to the public and government that they are “fit for purpose.”
The agency is looking for feedback on topics such as:
- What kinds of testing should AI development companies and their enterprise clients conduct regarding trust and safety;
- What kinds of data access is necessary to conduct AI audits and assessments;
- How can regulators and other actors incentivize and support credible assurance of AI systems along with other forms of accountability; and
- What different approaches might be needed in different industry sectors.
The insights gathered through the request for comment will help to inform the Biden administration’s ongoing work to ensure a comprehensive Federal government approach to responsible AI.
In October, the Biden administration issued the Blueprint for an AI Bill of Rights to help guide organizations on the development and deployment of AI, in an effort to help protect the rights of Americans.
However, the blueprint’s principles only serve as guidelines and are not enforceable law. The request for comment aims to identify concrete policies that can shape the AI accountability ecosystem.
Written comments are due to NTIA by June 10.