Senators ask Social Security for info on AI work

A bipartisan pair of senators on Wednesday requested information from the Social Security Administration about its use of artificial intelligence and urged increased safeguards if it is employed in benefit eligibility cases or the calculation of payment amounts.

Senate Finance Committee Chairman Ron Wyden, D-Ore., and ranking member Mike Crapo, R-Idaho, sent a letter to Social Security Administration Commissioner Martin O’Malley asking for information about how the agency is complying with the Biden administration’s March policy mandating safeguards for AI use. The move comes after the duo surveyed a trio of large federal departments about their use of AI last fall.

SSA has used AI in some form for more than two decades, mostly in the form of simple rule-based programs and predictive algorithms, to aid with tasks such as expediting disability claims that are likely to be granted and digitizing handwritten documents.

Though the lawmakers acknowledged that the agency has a history of using AI, they said the public focus on generative AI means agencies must engage with the public to assuage skeptics’ concerns.

“We also recognize the agency’s recent efforts to gather external input on its use of AI,” they wrote. “Whether incorporating newer technology like generative AI to improve customer experience and increase efficiency or leveraging predictive AI to provide decision support in the disability determination process, it’s critical that SSA meaningfully engage stakeholders, including its customers and employees, throughout the process of deploying new technology.”

Wyden and Crapo also warned O’Malley against overly relying on artificial intelligence as he seeks to chart a path out of Social Security’s customer service crisis, spurred primarily by a slow disinvestment in the agency by Congress, leading to the agency having its smallest workforce in 30 years amid the greatest number of beneficiaries in history.

“Notwithstanding its capabilities, AI is not a panacea for all challenges facing SSA,” they wrote. “Without the proper structure and guidelines for the procurement, deployment and monitoring of AI systems, SSA’s use of AI could reduce the effectiveness of its benefit administration processes, exacerbate improper payments and jeopardize beneficiaries’ financial security. To mitigate the risks associated with deploying AI across its programs, SSA must have strong governance frameworks in place that, among other important aspects, clarify the role of human discretion.”

Wyden and Crapo asked the SSA to provide information about its agencywide governance procedures, specifically how the agency is working to comply with the Office of Management and Budget March policy document establishing safeguards for the technology. They also asked for information about the AI systems employed in the agency’s disability determination process, the metrics measuring their success, and any safeguards to ensure the technology is used responsibly.

In a statement, Social Security Administration spokesman Mark Hinkle said the agency is committed to complying with all rules governing the use of artificial intelligence.

“The Social Security Administration is committed to the safe, secure and trustworthy development and use of AI, in line with the president’s executive order and OMB guidance,” he said. “We will continue to work with Congress and other stakeholders to ensure that we are advancing AI governance, talent and tools in ways that will help us deliver better for the American people."