AI Can Solve the Fiscal Crisis for Cities—If We Let It

American cities are teetering on the edge of a fiscal cliff. Covid-19 hammered municipal budgets: remote work displaced jobs, commercial rents plunged, and business travel and tourism collapsed, eroding key revenue sources. At the same time, heavy recovery subsidies and high inflation drove up expenses. The outlook remains grim: rising pension obligations, growing debt service, and escalating public works costs loom ahead. Unsurprisingly, most of the nation’s largest cities ran deficits in 2024, while new municipal debt issuance hit a record $513 billion, according to an industry trade group.

Cities have responded with a familiar playbook: hiking taxes on top- and middle-income earners while cutting services, from policing to Sunday library hours. These steps might suffice if the deficits were cyclical. But they fall far short of addressing the deeper structural crisis. Without a major efficiency breakthrough, much more painful reforms are inevitable.

Fortunately, a potential transformation has arrived in the form of artificial intelligence. Thanks to hundreds of billions of dollars in recent investment—spanning research, development, and infrastructure—AI capabilities have advanced rapidly, especially over the past five years. While some predict the advent of artificial general intelligence—AI capable of performing a very broad spectrum of human cognitive tasks—or even superintelligence, today’s commercially available tools are already capable of delivering the performance gains that cities need to avoid long-term financial instability. Off-the-shelf tools can improve student learning, enhance public safety, and boost government efficiency. With careful but swift adoption, AI can expand and improve service delivery, reduce costs, and help close chronic budget deficits afflicting major cities. In an estimate published in May, Boston Consulting Group said that 35 percent of government labor costs could be saved through AI technologies—a once-in-a-generation productivity improvement.

But entrenched interests are resisting. In Washington State, for example, Democrats in the state senate have introduced legislation that would effectively give public-sector unions veto power over AI. The bill would make adoption of the technology “bargainable” if hours, wages, or other employment terms are affected. Translation: good luck achieving needed reforms.

New interest groups have emerged, too, seeking to tie AI adoption to sweeping ideological goals. The Algorithmic Justice League argues that justice requires preventing AI from reinforcing “long-standing patterns of injustice” such as racial profiling or gender bias. Consulting firm Mercer similarly insists that AI, when “paired with a keen DEI lens,” can dismantle historical bias and advance organizational equity strategies.

Meantime, AI’s labor-market disruptions will drive some voters toward political leaders offering immediate—but hollow—relief. Protests have already called for banning AI in entire industries to protect redundant jobs. As with the long fight over occupational licensing, we’ll see a battle between those seeking broad public benefit and those defending their turf against change.

AI is not a jobs-protection mechanism or a vehicle for cultural crusades. It is a vital tool for mayors and city leaders to confront structural fiscal challenges. Done right, it could reshape urban life—replacing decline with dynamism and opportunity.

That will require bold leadership. Officials must anticipate backlash and act now: build AI-ready workforces, reimagine human services for quality and efficiency, and modernize city operations. Only by embracing the future can cities escape the stagnation of the present.

Aside from a few existential-risk believers who fear that an AI bot will take over the planet, the most common concern about AI is its potential to destabilize an already precarious labor market. Automation has long driven human progress, but it has rarely targeted well-educated professionals. AI upends this pattern; now, top workers worry that the machines are coming for their well-compensated jobs.

Empirical research supports these concerns. Studies by McKinsey, Deloitte, MIT, and others have found that a wide range of skills can already be automated using existing AI technologies. No two jobs are identical, making it hard to predict which specific roles will be displaced—but at the aggregate level, software engineers, paralegals, accountants, telemarketers, managers, translators, designers, and more face serious risk of rapid disruption, if not outright disappearance.

Given the relatively high pay of many of these professions, AI is expected to have a major impact on future income-tax receipts. A Brookings Institution analysis published earlier this year identified high-income metros like Washington, D.C., San Jose, and Seattle as being most at risk of AI-driven disruption. Other studies have pointed to metros like Tampa Bay that have historically absorbed back-office work from higher-cost cities.

As with all major technological shifts, these professionals will see some of their tasks eliminated by automation before the broader benefits of AI are fully realized. Blowback is inevitable. That makes it all the more urgent for city leaders to build a resilient, AI-ready workforce, equipped to maximize AI’s potential while minimizing its disruptions.

Cities should begin with a comprehensive review of their local labor markets. Using existing economic development and workforce staff—and investing strategically in new data tools—they must identify the types of work that professionals are doing and, crucially, which tasks are most vulnerable to automation.

These studies must go beyond job-level aggregates and focus on the more granular unit of tasks. Few jobs will be entirely displaced by AI, as most roles include tasks that can’t be fully automated. More important, many jobs could become significantly more valuable if low-productivity tasks become automated. Accurately gauging AI’s impact requires a far more nuanced analysis than traditional workforce studies. This research will likely be more ethnographic than statistical, aimed at understanding worker behavior and how it will evolve in response to AI.

Building on this research, city audit authorities should conduct a rigorous analysis of income-tax sources to assess the potential risks and opportunities that AI poses for future revenue. Some workers will become significantly more productive—and better compensated—while others may be rendered redundant. No city can navigate the coming transformation without a clear picture of its financial future. In some cases, little may need to change; for example, some researchers suggest that services-heavy cities like Las Vegas are relatively insulated from AI disruption (though others disagree). Either way, cities must understand the evolving shape of their workforces and revenue streams. They should collaborate to share best practices and the latest insights.

With a clearer understanding of their workforces and how they can adapt to AI, cities must develop workforce strategies that go beyond the generic recommendations typical of such reports. The rise of productive AI is a generational shift in the structure of the economy, with vast effects across nearly every domain of work. City leaders will be on the front lines of managing this transition and will need carefully to balance trade-offs based on their regions’ unique strengths and needs.

For instance, a ski resort town like Aspen, Colorado, may conclude that little needs to change in its workforce, since AI isn’t likely to affect the fundamentals of its outdoor recreation economy, which creates an estimated 40 percent of all local jobs. Even in fields facing heavy automation, such as medicine, retirement communities like The Villages, Florida, or Sun City, Arizona, may find that AI-driven gains in physician productivity improve seniors’ health more than they threaten medical employment. In fact, such improvements could increase demand for medical services.

On the other hand, economically balanced cities like Minneapolis may face a more complex path. The city has strengths in scientific research and medical devices, as well as in finance, corporate retail, and professional services. Traditionally, this diversity has helped Minneapolis weather downturns in any one sector, but AI threatens to disrupt all of them at once. A strong strategic plan will need to identify overlaps in these disruptions while ensuring that each industry boosts its global competitiveness through productivity gains.

While attention will focus on AI eliminating entire jobs, the more daunting challenge will be the great number of roles made partially redundant—prompting organizations to consolidate their workforces. Klarna CEO Sebastian Siemiatkowski said in May that the high-flying fintech company has downsized 40 percent, from 5,000 to 3,000 employees, through AI automation in areas like customer support. “If you go to LinkedIn and look at the jobs, you’ll see how we’re shrinking,” he told CNBC. CEOs like him will ask whether they still need 100 accountants in the tax department or just ten highly trained, highly compensated professionals to oversee more automated systems. A well-designed strategic plan should account for these nuances and offer clear steps to build employment resilience.

With so much disruption ahead, pressure will mount to regulate—or even prevent—the automation of many professions. Political leaders will seize the mantle of job protection rather than of championing innovation. Earlier this year, California State Senator Jerry McNerney introduced the No Robo Bosses Act, warning that “there are currently no safeguards to prevent machines from unjustly or illegally impacting workers’ livelihoods and working conditions.” The bill is backed by the California Federation of Labor Unions.

These voices should be heard and engaged—but rejected. Automation promises immense benefits that can strengthen the economy’s long-term outlook. No group should be allowed to immunize itself from necessary change at the public’s expense. And in the end, urban competition will reward cities that adapt early. The future is coming, whether voters like it or not.

For city leaders, executing a clear strategy with strong public-private partnerships will be essential. They must consistently communicate that the workforce can adapt and become AI-ready and that rising efficiency offers a path not only to fiscal stability but to a higher quality of life for all residents. Fortunately, corporate leaders and individual workers alike have strong incentives to adapt. Their entrepreneurial drive should be encouraged, not constrained.

City budgets are dominated by spending on human services, including education, health care, public safety, and welfare. These services are labor-intensive—and therefore extraordinarily costly to provide. AI offers a way to improve service quality significantly, while containing current and future cost growth, but only if entrenched interests can be overcome through a sustained focus on modernization.

Take education. America’s public schools are a study in dysfunction. Literacy and numeracy scores on major international tests have plummeted since Covid-19, sometimes to their lowest levels on record. Schools remain stuck in a model built during the industrial age, designed to process large numbers of children into workers as efficiently as possible. Technology, when used, is often bolted onto traditional teaching rather than integrated into it—contributing to teacher burnout and persistently poor student outcomes.

Education research consistently shows that individualized tutoring is one of the few interventions that reliably improve student performance. AI, then, has the potential to play a transformative role in schools. Current tools can design customized lessons based on each student’s past performance, and it’s already possible to provide every student in the country with a customized AI tutor. If the research holds up in practice, the positive effects could be extraordinary.

Using AI in the classroom doesn’t mean that teachers are now obsolete or that students will be herded into dystopian rooms to interact all day with glowing screens and computer-generated tutors. Rather, AI will complement teachers, giving them more time to help students develop metacognition: the ability to understand and direct their own thinking. With more productive, individualized learning, students can skip redundant homework on material they’ve already mastered and gain flexibility in the school day for creative pursuits and recreation. This shift could even help counter the rise in attention-deficit disorders seen over the past three decades.

That’s a significant opportunity for education if we can seize it. The challenge will be less about job displacement and more about skill retraining. The National Education Association, the nation’s largest teachers’ union, stated last year that “AI-enhanced technology should aid educators, but it cannot and should never aim to replace them,” insisting that educators be involved at every stage of AI adoption to “guarantee that these tools are used to improve job quality and enhance performance.” Though teachers and even their unions have welcomed technology into classrooms, such statements make clear that job protection remains the first priority.

But AI will push education leaders fundamentally to rethink schooling. As Rebecca Winthrop of the Brookings Institution said recently, “I’m very excited about AI because it has to move us from the age of achievement into the age of agency, as I call it, where you could have schools break open that sorting and ranking and really bring much closer together knowledge acquisition with knowledge application.” The key is to take on this transition directly, rather than cling to a model that no longer keeps pace with the needs of society.

Early strategic investment can shift the narrative, promote the spread of the best approaches, and foster collaboration in the pursuit of greater efficiency. Providing clear pathways for the teaching profession to embrace and realize fully the benefits of new AI technologies will help minimize resistance. When necessary, leaders must push back vigorously against efforts to block progress in defense of the status quo. America’s children deserve the best—especially after years of falling short. AI offers a vital opportunity to improve their outcomes at scale and speed.

Public safety presents even more complex challenges. AI could dramatically boost policing productivity—streamlining paperwork, analyzing CCTV footage, and processing evidence like rape kits, where backlogs remain severe. As with teaching, AI promises major gains here, but few expect the beat cop to be replaced by a robot anytime soon.

Yet activist organizations are already pushing for major constraints on the use of these tools. Many of their concerns are valid; no citizen in a democracy wants to live under an AI-equipped Orwellian police state. But some proposals go well beyond reasonable safeguards, erecting a thicket of rules that no AI system could realistically navigate. In many cases, the goal is less about public protection than about blocking policing’s future.

Once again, city leaders will need to bridge the gap between the public’s demand for improved public safety and activist opposition to innovation. There is precedent. In the early 1990s, the NYPD’s development of its CompStat system combined improved digital recordkeeping with redesigned organizational processes to align policing priorities more effectively with data. The approach proved so successful that police departments across the U.S. and around the world have adopted it.

Police officers—like most professionals—are creatures of habit, shaped by the administrative procedures that structure and constrain their work. Early investment in leadership and training on the capabilities of new AI technologies, paired with encouragement to adopt time-saving tools, can boost performance and reduce institutional resistance.

Education and public safety are not exceptions. AI will bring similar patterns of rapid modernization and institutional pushback across all human services. The public stands to benefit enormously from more effective government. City leaders must prioritize those benefits over the preferences of professionals unwilling to adapt.

Perhaps the hardest part of the AI transformation will involve the legions of office workers who staff city bureaucracies. Unlike the roles of teachers and police officers, many administrative positions rely on skills expected to be replaced entirely by automated tools already available on the market.

The benefits of automating administration are obvious. As almost anyone who interacts with city government knows, filling out the correct forms and acquiring the right licenses and titles can require many stops across multiple departments. In big cities like New York, these processes are so byzantine that an entire industry of so-called expediters exists just to manage them, at a significant cost to anyone buying a home or starting a new business.

In extreme cases, city government workflows—rezoning is an egregious case—can take weeks or even months to complete as data are compiled and multiple approvals are secured. With AI, that same process could occur in seconds. As with the rise of 311 systems two decades ago, these procedures could be streamlined into a single, user-friendly interface for public access. Today’s enterprise AI tools are built to navigate siloed databases and answer complex queries with proper access controls; there’s little reason government should be any different.

This transition will have a profound impact on city performance—and on municipal workforces. The Bureau of Labor Statistics estimates that there are roughly 15 million local government workers, the most in history. Automation in government services could displace millions of them, and those affected are likely to resist such changes. AFSCME, the nation’s largest municipal workers’ union, stated in its AI policy last year that it “will advocate to ensure that labor has a meaningful, decision-making role on the use of AI in the workplace through collective bargaining at every stage of the process, including design, implementation, and monitoring of AI systems.”

One of the most powerful arguments that unions and other interest groups make against AI systems is that they are unfair, biased, and discriminatory. But as I noted in City Journal last year, such charges are routinely leveled at human-made decisions in government, too. (See “United States of Algorithms,” Summer 2024.) The key is to scaffold both human and AI systems with proper due process, ensuring that even the fastest AI-driven decisions remain fair.

City leaders must ensure that new AI systems get implemented in ways that deliver the operational gains essential to healthier budgets and better services. That will be especially challenging in cities where unions are major political donors and where competitive pressure—so powerful in the private sector—exerts far less influence. Overcoming these headwinds will require strong public-private partnerships and a clear, compelling message about the broad benefits to all citizens.

Like any new technology, AI is no panacea. It cannot simply be dropped into a city and expected to solve deep-rooted problems. Its impact will depend on how well it integrates with existing systems, the quality of training for government employees, and thoughtful outreach to the public. Without early investment and a clear strategy, cities risk getting left behind in the rush toward the future.

Implemented thoughtfully, though, AI offers a rare opportunity to improve life for all urban residents. That’s especially valuable as cities confront mounting debt and retirement obligations and the urgent need to restructure budgets for long-term sustainability. The shift to more automated, responsive governance won’t be easy—and it will face loud opposition. But American cities still have the chance to step back from the edge of financial crisis and reclaim their role as models for the world. That’s a future worth fighting for.

Top Photo: Malte Mueller / fStop via Getty Images