Data tracking, AI and modernization: How SSA has ditched its paper-based past

For more than two decades, the Social Security Administration has worked to shed its well-earned image as a “paper-based agency,” doing everything from shifting away from hard copy signatures to embarking on a large-scale modernization plan.

Now armed with a robust artificial intelligence use case inventory and a splashy new data-tracking tool, SSA’s technology leaders feel especially bullish about where they stand in their digital journey.

“There’s a lot of cultural, organizational change that’s going on internally to really bring everybody to the forefront and really understand that the customer really needs to drive how we deliver [and] what we deliver,” said Marcela Escobar-Alava, the SSA’s chief information officer.

In an interview with FedScoop, Escobar-Alava and two other top SSA officials spoke of the ongoing digital transformation within the agency that has accelerated thanks to what they say is an embrace of data and artificial intelligence — and a sense of urgency given a looming change in leadership.

SSA Commissioner Martin O’Malley was sworn into his post at the end of 2023, taking over for acting Commissioner Kilolo Kijakazi on a term that will end in January. With that truncated tenure in mind, O’Malley hit the ground running on a project near and dear to his heart: SecurityStat, which launched in February.

Based on the New York City Police Department’s CompStat crime data model, SecurityStat measures and visualizes various performance metrics critical to the 70 million people reliant on Social Security benefits and the millions more who interact with SSA services. 

O’Malley had previously adapted the NYPD’s tool for Baltimore during his time as its mayor and for Maryland when he served as governor. Though neither program thrived after O’Malley left office — and CompStat, CitiStat, StateStat and other programs of their kind have complicated legacies — SSA struck the new commissioner as an ideal venue for a stat-based approach to governance.

“This agency is absolutely built for that sort of performance management routine,” said Dustin Brown, SSA’s chief operating officer and acting chief of staff. “We’re already seeing pretty impressive improvements based on the SecurityStat approach.”

It took the agency’s tech staff about two months to get the program’s website up and running. The first image users now see is a graphic that “illustrates the widening gap between staffing levels and the growing number of customers,” a point Brown said was especially important to note given that SSA is operating at a 27-year staffing low while having to serve more beneficiaries than ever before.

Having the “lowest levels of employee engagement of any major agency in the federal government [and] the most exhausted workforce in the federal government” leads to “major operational challenges,” Brown noted. 

But those staffing hurdles seemingly didn’t deter O’Malley. At launch, Brown said the commissioner instituted two-week check-ins as opposed to looking at metrics on an annual basis. The data selected by SSA to highlight on the page represents some of the agency’s most pressing customer service challenges: wait times for the national 800 number, processing time for benefits, and disability determinations. 

For a “data rich” agency like SSA, Brown said it made sense to O’Malley to give website users the ability to “see that granular data.” The page serves the purpose of providing SSA’s workforce with a “better understanding” of various metrics as well as “where we may need to focus our attention,” he added.

While those trouble areas can now be visualized more clearly, the next step for SecurityStat is adding open data and visualization dictionaries to the site. Brian Peltier, the agency’s chief artificial intelligence officer and deputy CIO, said those integrations will allow users to “take a look at what’s going into these numbers and why we are coming to these conclusions.”

As the agency works to implement those open data and visualization dictionaries to SecurityStat this summer, other IT work has continued unabated. Escobar-Alava mentioned a host of ongoing digital priorities: improved digital notifications, more services and information for users logged in to mySSA, and internal training sessions intended to stem the tide of workforce departures and generally “simplify the lives of technicians.”

More broadly, Escobar-Alava is head down in formulating a digital modernization strategy for the agency, a process that involves a lot of listening to glean “insights into where we should be focusing our resources or energy or investments.” Given the SSA’s history of building “lots of monolithic applications” in its paper-based era, Escobar-Alava said there’s been a lot of work done to strip out elements of legacy tools and make sure that the target architecture is established and communicated to all IT stakeholders across agency offices.

Ultimately, Escobar-Alava said she and her team are working on “ensuring that the technology organization at large can start to shift how we deliver” digital services overall.

Much of that broader shift falls under Peltier’s purview leading the agency’s AI work — a topic that IT leaders at SSA have thought a great deal about. The agency comes to the AI revolution from a position of strength, with forms of the technology in use at SSA for 20-some years, according to Peltier, and its AI work characterized in some circles as pioneering. 

“The more recent boom of artificial intelligence is around generative AI,” he said, “but we’ve been using it for many years before that.”

Of the SSA’s 14 current AI use cases, Peltier pointed to its Intelligent Medical Language Analysis Generation tool, better known as IMAGEN, as one of the agency’s biggest successes. The product leverages machine learning technologies and predictive analytics to streamline disability determinations, helping SSA’s workforce more effectively unearth customer impairments and speed up the decision-making process.

Other tools employed by the agency use predictive AI models to improve efficiencies, including one that flags fraudulent claims and another that screens disability applications and identifies when a favorable determination is especially likely.

“Overall, it’s really helped us move more of these [disability] cases faster … and more effectively,” Peltier said, noting a 157% year-over-year jump in process rates and plans for expanded use of the tool across SSA teams.

Sustaining the agency’s digital momentum will depend in part on Congress. Brown said O’Malley has been meeting regularly with lawmakers to make the case for President Joe Biden’s fiscal year 2025 budget request and the 8% boost for SSA that comes with it. 

Those resources, Brown said, would help increase agency “staffing levels so that we can do a much better job in getting those initial determinations much more quickly to individuals,” an increasingly critical task since SSA’s backlog is over a million individuals and “could go up significantly” absent a funding infusion. 

Regardless of what happens in current and future budget fights, Brown said agency leaders want “everyone” at SSA to think about data in a “much more holistic kind of way” — an approach that resonates with Escobar-Alava especially.

As the agency flips the switch from paper to digital and fully embraces AI, modernization, data-focused projects like SecurityStat and other initiatives, Escobar-Alava said IT leadership is aligned with Brown and the rest of the executive team on its overarching tech strategy.

“Now we need to start connecting the dots with what is really driving the business measures and some of the results,” she said. IT leaders are “very focused on bringing all of these pieces together to deliver a much better experience for the public and our employees.”