Federal Agencies Unveil AI-Powered Future Running Entirely on Software That Predates the iPod

WASHINGTON — The federal government unveiled its most ambitious technology modernization initiative in a decade Wednesday, promising to harness the power of artificial intelligence, machine learning, and next-generation data infrastructure to deliver faster, smarter, more responsive public services to every American.

The underlying systems processing those services run on COBOL.

The Federal Digital Transformation Initiative, announced at a press conference featuring a professionally designed logo, a 90-second video with an uplifting soundtrack, and a slide deck that used the phrase "paradigm shift" eleven times by the third slide, represents what the Office of Management and Budget called "a generational leap forward in how government serves citizens."

A footnote on page 47 of the accompanying technical documentation confirmed that the leap would be executed using a database management platform whose manufacturer ceased operations in 2002 and whose last security patch was issued during an administration that no longer exists.

"We are absolutely committed to bringing federal technology into the 21st century," said Deputy Chief Information Officer Raymond Blass at the podium, standing in front of a screen displaying a rendering of a gleaming digital interface that bore no resemblance to any system currently operated by the federal government. "This initiative represents a fundamental transformation in our technical posture."

He did not take questions about the technical posture.

The Infrastructure Beneath the Infographic

America's federal technology ecosystem is, by the government's own accounting, a geological formation — layer upon layer of systems deposited across decades, each built atop the last because removing the previous layer proved either too expensive, too complicated, or too likely to accidentally stop processing Social Security checks.

The Government Accountability Office has identified more than 10,000 federal IT systems currently classified as "legacy," a word that in technology contexts means "functional in the way that a 1987 pickup truck is functional: it runs, you just can't get parts for it anymore." The oldest of these systems were built before the commercial internet existed. Several predate the fax machine's transition from novelty to standard office equipment — which, given current government practice, makes them roughly contemporaneous with the fax machine's current status as government standard office equipment.

The Social Security Administration processes trillions of dollars in payments annually on systems written in COBOL, a programming language developed in 1959 that the average computer science graduate has never encountered and would not recognize. The IRS runs core tax processing on infrastructure so old that internal documentation reportedly describes certain components as "mature," which is the bureaucratic equivalent of describing a milk carton as "characterful."

"Legacy modernization is a priority," said a senior technology official who agreed to speak on background, meaning they agreed to speak only if nothing they said could be traced back to them or their agency or their general area of the building. "It's been a priority for about twenty years. We're very committed to it."

The AI Strategy, Annotated

The Federal Digital Transformation Initiative's artificial intelligence component is, on its surface, impressively ambitious. The initiative proposes to deploy AI-assisted processing across fourteen major federal agencies, enabling faster benefit determinations, more responsive citizen services, and — in language lifted almost verbatim from a consulting firm's proposal — "intelligent automation of high-volume transactional workflows."

The fine print is instructive.

Phase One of the initiative, covering the first eighteen months, involves "assessment and architecture planning" — which, translated from government technology procurement language, means hiring a contractor to look at the existing systems and produce a report about how complicated they are. The report is expected to cost between $8 million and $12 million and to conclude that the systems are very complicated.

Phase Two, covering months nineteen through thirty-six, involves "interface layer development" — meaning the AI tools will be built to sit on top of the existing legacy systems rather than replacing them, communicating with 40-year-old databases through a translation layer that experts describe as "technically feasible" and that one senior engineer, speaking with the weary candor of someone who has watched three previous modernization initiatives collapse, described as "bolting a Tesla dashboard onto a horse."

Phase Three, covering full system modernization and legacy decommissioning, is projected to begin in fiscal year 2028. It has been projected to begin in fiscal year 2028 since 2021, when it was projected to begin in fiscal year 2026, having previously been projected to begin in 2024.

"The timeline is aggressive but achievable," said Blass, when a reporter who had read the footnotes asked about Phase Three. "We're very confident in our roadmap."

The roadmap does not include a projected completion date. It includes a projected date for producing a projected completion date.

Vendors, Very Excited

The technology industry's response to the initiative was swift, enthusiastic, and almost entirely focused on the contracting opportunities.

Several major consulting firms issued statements praising the government's "bold vision" and noting their extensive experience in exactly this type of transformation — experience that, in at least two cases, includes having been the prime contractor on a previous federal modernization initiative that did not achieve its stated objectives, was restructured twice, and was quietly declared complete at a scope representing approximately 30 percent of the original mandate.

"We are incredibly excited to support this initiative," said a spokesperson for one firm, which Officially Absurd has agreed not to name in exchange for absolutely nothing, because they didn't ask. "Our team brings deep expertise in federal IT modernization, legacy system integration, and helping agencies navigate complex technical environments."

The firm's federal IT division was founded in 2018. The systems it will be navigating were built before its founding partners were born.

What Citizens Can Expect

For ordinary Americans, the initiative promises a transformed experience of interacting with federal agencies — eventually. In the near term, officials say, citizens should expect the same experience they currently have, delivered through a new web portal whose front end will be modern, responsive, and attractively designed, and whose back end will be querying the same database it has been querying since the Clinton administration.

The IRS confirmed that its new AI-assisted tax filing interface, projected for limited rollout next year, will reduce average processing times by up to 40 percent for straightforward returns. For complex returns, it will refer users to a PDF form, which must be printed, completed in black ink, and mailed to an address that is different from the address printed on the form you downloaded last year.

The Social Security Administration noted that its modernization timeline would not affect current beneficiaries, who will continue to receive payments processed by systems that have been running without interruption since before most of their grandchildren were born, and which the agency describes, with genuine and justified pride, as "extraordinarily reliable."

"We're not dismissing the old systems," said one technology official. "In many ways, they're the foundation of everything we're building. You don't tear down the foundation."

Experts in federal IT modernization say this is technically accurate and also precisely the problem.

Full modernization, officials confirmed, remains on track for completion by a date that will be determined following the completion of Phase One, which will establish the parameters for determining Phase Two, which will inform the scheduling of Phase Three, which will, at that point, be projected to begin in approximately 2031.

The logo, at least, is very good.