New AI Regulations Imposed on Government Agencies

( – The White House has released new rules on how federal agencies can use Artificial Intelligence (AI) computer technology. The binding guidance came through a memo from the Office of Budget and Management.

The rules address the need to protect users, agencies, and citizens from problems with the powerful technology while also using it to the best advantage. But there are few specific details about how government agencies will do this.

For example, the rules mandate that federal agencies appoint a “chief intelligence officer” to supervise the roll-out of AI technology in the agency. These protocols must accomplish testing, monitoring, and evaluating the effects of AI technology, but there are no detailed instructions on how to do this yet.

Agencies are also required to have safeguards in place by December 1st, but it is not clear what form those safeguards would take, or what specific steps supervisors would need to take to prevent unnecessary risk. However, agencies that don’t have these safety protocols in place would be required to stop using AI until they do.

The federal government says it plans to hire 100 professional experts in AI by the summer of this year.

Vice-President Kamala Harris introduced the guidance in a press call on March 28th, and her remarks suggest the Biden Administration is tightly focused on issues of alleged bias and discrimination in AI. She said agencies will have to be transparent about their use of AI and will have to “verify” that the technology does not “endanger the rights and safety” of citizens.

The hypothetical example she chose to illustrate this involved a scenario at the Veterans Administration. If the VA were to use AI to diagnose illness, she said, the VA would have to show that the technology “does not produce racially biased diagnoses.”

That may be a tall order if the AI systems the government uses are at all similar to Google’s Gemini AI. When the company introduced the technology in February, experts and users alike decried and mocked the AI for creating historically inaccurate images of great figures. User requests to create images of Vikings, for example, resulted in depictions of black African or East Asian faces in Viking costumes.

Copyright 2024,