Senate Democrats ask OMB for more regulation, support to mitigate algorithm bias

Two Democratic senators have asked the Biden administration to do more to reduce the risk of artificial intelligence algorithms making biased decisions.

Senators Edward Markey, D-Massachusetts, and Majority Leader Chuck Schumer, D-New York Times Democrat, said in a letter Monday to Office of Management and Budget Director Shalanda Young that federal agencies must establish more safeguards to prevent algorithmic discrimination.

“Without new safeguards, today’s supercharged AI-based algorithms risk amplifying and compounding the discrimination that marginalized communities already face because of poorly trained and tested algorithms,” the letter reads. “The stakes—and the harms—are especially high when actors use algorithms to make ‘important decisions,’ like whether someone applies for a job, gets hospital treatment, is accepted into an educational institution, or qualifies for a mortgage.”

The senators want federal agencies that use AI technologies in their operations to be required to develop safeguards and build capacity around civil rights protections around AI. Algorithmic discrimination has already drawn federal attention, most notably when the Department of Housing and Urban Development issued a warning in May of this year to draw attention to how AI-powered application software could wrongly discriminate against applicants.

Markey and Schumer note that the Biden administration has also taken “significant steps” to limit the scope for algorithmic discrimination, primarily through guidance in President Joe Biden’s executive order on artificial intelligence.

They noted that OMB, in particular, has been at the forefront of many of these orders, as well as the agency’s most recent policies and guidance on AI, particularly regarding “artificial intelligence affecting rights.”

The Senators recommend that the Office of Management and Budget (OMB) provide AI Chiefs with adequate resources and expertise to mitigate the risks that AI algorithms pose to civil liberties, and create and fund civil rights offices in agencies that use AI to make decisions where they do not currently exist.

“These new offices — similar to the existing civil rights offices — should be staffed with technologists and experts in algorithmic discrimination whose job responsibilities include mitigating algorithmic bias and discrimination and facilitating proactive and sustained outreach to civil rights stakeholders and affected populations,” the lawmakers wrote.

In addition to increasing guidance and staffing to address harms from algorithmic use, the senators asked OMB to provide evidence that customers of federal government services can opt out of AI-based algorithms, a move that has been seen with AI-based algorithms used in biometric technologies.