The views expressed are those of the author and do not necessarily reflect the views of ASPA as an organization.
By Md Eyasin Ul Islam Pavel
October 16, 2026
In February 2026, a veteran at the Washington, D.C., VA Medical Center called to schedule a mental health appointment. According to a Vet Voice Foundation report, the VA’s own website listed the wait at seven days. The facility told him approximately four months and said longer-term therapy was no longer available. This is not an isolated case. A Vet Voice Foundation study tracking 21 VA medical centers found that since recent rounds of staffing cuts, wait times increased at 71 percent of facilities studied. These are numbers from a system where physicians already face significant administrative burdens. According to the American Medical Association, reporting in August 2025, documentation and administrative tasks continue to consume substantial amounts of physicians’ time, time that AI could help free up to serve veterans currently going without care.
Washington celebrates efficiency. Veterans experience waiting.
This is the capacity crisis AI could actually help solve.
The more important question is this: What could America accomplish if AI gave public servants more time to serve the public?
The emerging evidence suggests that AI’s greatest contribution may not be replacing workers at all. It is relieving them of routine work. The VA says early AI pilots are already helping staff save time. Some tools, such as VA GPT, Teams Premium and GitHub Copilot, have saved employees up to 10 hours each month. The department says AI can reduce administrative burdens and give staff more time for direct, high-impact services. The Congressional Budget Office notes that many economists see AI as having the potential to boost productivity and economic growth, while the Government Accountability Office has documented how federal agencies are already using AI to streamline administrative work and improve public service delivery.
Those findings point to something more valuable than efficiency.
They point to capacity.
Efficiency saves resources. Capacity expands what institutions are able to accomplish. That distinction changes the entire conversation.
America does not suffer from a shortage of public problems. It suffers from a shortage of institutional capacity to address them.
According to the 2024 National Survey on Drug Use and Health, millions of adults with mental illness received no treatment in 2024. Meanwhile, the aging population will place growing pressure on caregiving systems already stretched to their limits. These are not isolated failures. They are symptoms of the same underlying constraint: Government routinely has more responsibilities than it has time and people to fulfill them.
Artificial intelligence has the potential to change that equation.
Imagine a social worker who spends hours each day completing paperwork, writing case summaries and documenting routine interactions. If AI reduces that administrative burden, the default response should not be eliminating a social worker’s position. It should be asking how those newly available hours can be used to serve more people.
More home visits to vulnerable families. More veterans connected to mental health services. More outreach to people experiencing homelessness. More inspections of unsafe housing. More attention devoted to preventing crises before they become emergencies.
The same employee. The same salary. A greater public impact.
The same principle extends beyond government. Businesses and nonprofit organizations also devote enormous resources to repetitive administrative work. When AI creates additional capacity, organizations can choose to invest that time in innovation, customer relationships, workforce development and community partnerships instead of treating every productivity gain as an opportunity to reduce payroll.
This is what policymakers should recognize as AI’s capacity dividend: the additional public value that becomes possible when technology frees people from routine work and those recovered hours are reinvested in solving problems that have long exceeded society’s institutional capacity.
Whether we realize that dividend is not a technological question. It is a policy choice.
This is not an argument against efficiency or fiscal discipline. Governments should absolutely use AI to eliminate unnecessary paperwork and outdated administrative processes. But productivity gains should not automatically become headcount reductions. Before agencies eliminate positions, they should first ask whether those newly available hours could produce greater public value elsewhere.
Congress should require federal agencies to file AI reinvestment plans documenting how productivity gains will expand public services before any position is eliminated.
Federal agencies should identify where additional capacity is most urgently needed, explain how employees can be redeployed into those roles and, where necessary, provide targeted training so they can succeed in their new responsibilities. Success should no longer be measured only by dollars saved or positions eliminated. It should also be measured by outcomes: more families receiving food assistance, more seniors receiving in-home support, more domestic violence victims accessing shelters and legal aid quickly and safer communities.
For decades, policymakers have asked how government can do more with less. Artificial intelligence gives us an opportunity to ask a better question: How can government do more for more people with the capacity it already has?
That is the real promise of AI. Not simply a smaller government. A more capable one.
A VA clinician who spends three hours on paperwork could be seeing more veterans if AI helped reduce that burden. AI can give those hours back. The only question is whether Washington will let clinicians use them.
Author: Md Eyasin Ul Islam Pavel is a Ph.D. candidate in public affairs at the University of Texas at Dallas, where his research focuses on federal workforce policy, public management, artificial intelligence governance, public sector innovation and inequality. His work has appeared in PA TIMES. LinkedIn profile.