The Teacher Shortage Trap: How AI in Education Becomes a Substitute for Paying Teachers

A public school teacher in the United States earns, on average, 26.6 percent less than a worker with comparable education and experience in another profession. That number comes from the Economic Policy Institute's analysis of 2023 wage data, and it has been growing for three decades. In 1996, the penalty was roughly 6 percent. It has more than quadrupled since.

The same week that data point circulated through education policy circles, a humanoid robot named Figure 03 walked into the White House. First Lady Melania Trump introduced it to an audience of international first spouses and proposed that machines like it could serve as educators. "Imagine a humanoid educator named 'Plato,'" she said. "Access to the classical studies is now instantaneous."

The gap between the problem and the proposed solution is the story. The teacher shortage is not a technology problem. It is a compensation crisis with a technology alibi.

The 27 Percent Discount

The Bureau of Labor Statistics tracks what Americans earn by occupation. In May 2023, the median annual salary for an elementary school teacher was $61,690. For a high school teacher, it was $65,220. These are jobs that require at minimum a bachelor's degree, often a master's, plus state certification and continuing education credits.

Compare those numbers to other professions requiring similar credentials. An accountant earned a median of $79,880. A software developer pulled in $127,260. The teacher, working with roughly the same level of formal education as the accountant, took home about $18,000 less per year.

The Economic Policy Institute has tracked this disparity as the "teacher wage penalty" since 1979. In inflation-adjusted terms, teacher pay has barely moved in four decades. Other professions with similar educational requirements have seen steady gains. The result is a penalty that widens every year, like a crack in a foundation that nobody repairs because the house is still standing.

This is not a regional anomaly. The penalty exists in every state, though it varies. In Colorado, teachers earn roughly 38 percent less than comparable workers, the widest gap in the nation. In Wyoming, the gap is smaller but still significant. No state has achieved parity.

55,000 Empty Classrooms

The Learning Policy Institute estimated that at the start of the 2023-24 school year, approximately 55,000 teaching positions sat vacant across the United States. Tens of thousands more were filled by underqualified substitutes or individuals teaching outside their certification area.

The shortage is not evenly distributed. Special education, mathematics, and science positions are the hardest to fill, because professionals with STEM credentials can earn substantially more outside education. Bilingual education positions remain chronically understaffed, a problem that grows as the English-learner student population increases.

Geography compounds the problem. Rural districts cannot offer the salaries that urban and suburban districts provide. High-cost urban districts, paradoxically, face similar difficulties because their salary schedules have not kept pace with local living costs. A teacher in San Francisco earning $65,000 confronts a housing market where the median one-bedroom apartment rents for over $3,000 per month.

The pipeline is drying up. Enrollment in teacher preparation programs declined by approximately a third between 2010 and 2020, according to data from the American Association of Colleges for Teacher Education. Fewer students choose teaching as a career because they can see the numbers. The shortage is self-reinforcing: as vacancies rise, remaining teachers absorb larger class loads and additional duties, accelerating burnout and attrition. Roughly 8 percent of public school teachers leave the profession each year. Not all retire. Many simply leave.

What Finland, Singapore, and South Korea Actually Did

The United States is not the only country that has faced a teacher shortage. Several countries solved theirs. None of them used robots.

Finland restructured its education system beginning in the 1970s. The core decision was to make teaching a high-prestige, well-compensated profession. Finnish teachers earn roughly 92 percent of what comparably educated professionals make, far closer to parity than most countries achieve. Admission to teacher education programs at Finnish universities has historically had acceptance rates around 10 percent, more selective than many medical schools. The profession carries social prestige that in the US is reserved for doctors and lawyers.

Singapore took a different route to the same destination. The government benchmarks teacher salaries against the top third of graduate starting salaries in the economy. The career ladder is steep and structured, with significant salary increases at key promotion points. The National Institute of Education is the sole pathway into teaching, maintaining quality control over the pipeline.

South Korea compensated through starting salary. A new Korean teacher earns roughly 1.1 times the national GDP per capita, compared to the OECD average of about 0.9 times. The teaching certification exam is competitive and carries prestige similar to the civil service examination. Teacher attrition in South Korea is among the lowest in the OECD.

The pattern repeats across contexts. The OECD's Education at a Glance reports consistently show that countries paying teachers above the median graduate salary report lower vacancy rates and higher application volumes. The mechanism is not mysterious. When a profession pays well and carries status, qualified people apply for it.

This pattern extends beyond wealthy nations. India's Kendriya Vidyalaya system, the network of centrally funded schools, pays teachers well above the national median salary and has no significant recruitment problem. But India's state-funded schools, where pay is far lower, face massive vacancies in rural areas. In sub-Saharan Africa, UNESCO projects a need for 15 million additional teachers by 2030, driven almost entirely by funding constraints rather than a lack of willing candidates. Brazil's Fundeb funding mechanism and its 2008 teacher pay floor legislation represent an attempt to apply the same logic, though implementation remains uneven. Chile's 2016 teacher salary reforms, which increased starting salaries and introduced performance bonuses, have begun to reverse declining enrollment in teacher preparation programs. The evidence base is global: pay teachers and they come. Underpay them and they leave.

The Alpha School Model

Alpha School represents the logical conclusion of a different approach. Rather than raising teacher pay to attract qualified professionals, Alpha School reduces the role of human teachers to minimize costs.

The model works like this: students spend approximately two hours per day on AI-guided academic instruction. The remainder of the day is devoted to activities supervised by "guides," a term that replaces "teacher" in Alpha School's vocabulary. These guides are not required to hold traditional teaching certifications. The AI handles the content delivery.

Tuition varies widely by location, from roughly $10,000 per year in Brownsville, Texas, to $75,000 in San Francisco. The model has drawn attention not for affordability but for its premise: that software can replace much of what a teacher does. No peer-reviewed studies of Alpha School's academic outcomes have been published. The company has not released longitudinal data on student achievement, college admission rates, or long-term performance metrics. What it has released is a cost structure.

Marc Beckman, a senior adviser to Melania Trump, has been instrumental in connecting education technology companies with the White House. An Alpha School student attended the 2026 State of the Union as the First Lady's guest. The school's model has received prominent visibility in White House education circles, even as independent verification of its academic claims remains absent.

The business model is transparent. Staff salaries and benefits account for roughly 80 percent of a typical school's operating budget. Any model that reduces teacher headcount delivers immediate and significant cost savings. Software has a marginal cost that approaches zero as it scales. A teacher's marginal cost is fixed per classroom, per year, per salary. Alpha School did not solve an education problem. It solved an expense problem.

The Executive Order Scaffold

In April 2025, President Trump signed an executive order titled "Advancing Artificial Intelligence Education for American Youth," establishing a task force to promote AI literacy, integrate AI into education, and train educators. The order did not mandate replacing teachers with AI. It did not reference the teacher shortage directly. What it created was infrastructure.

When a federal task force is chartered to promote AI integration in schools and coordinate with the private sector, the incentive structure is directional. Companies that can demonstrate cost savings will attract attention. Companies that can show a product replacing an expensive human function with a cheaper automated one will attract funding. The executive order is a scaffold, not a blueprint, but scaffolds determine the shape of what gets built on them.

Melania Trump is not a formal member of the task force. But her White House summit, occurring less than a year after the executive order, served as a public demonstration of the policy's intended direction. The First Lady stood next to a robot and told an international audience that "the future of AI is personified."

The policy pipeline is now in place: an executive order authorizing AI education initiatives, a White House platform promoting humanoid educators, and a private school model demonstrating the commercial viability of replacing teachers with software. Each piece supports the others. None of them addresses teacher pay.

The $55 Billion Question

The arithmetic of closing the teacher wage gap is straightforward. The National Center for Education Statistics counts approximately 3.7 million public school teachers in the United States. Bringing their average compensation to parity with comparably educated professionals would require an average raise of roughly $15,000 per teacher. That produces an annual price tag of approximately $55 billion.

That number sounds large in isolation. In context, it is modest. Total US K-12 education spending exceeds $900 billion per year. Adding $55 billion represents roughly a 6 percent increase. The US defense budget for fiscal year 2024 was $886 billion, and it increased by more than $55 billion in a single legislative cycle. Congress has demonstrated repeatedly that it can allocate this kind of money when it chooses to.

Meanwhile, the global AI-in-education market is projected to reach tens of billions of dollars annually by 2030, according to estimates from market intelligence firms. That money flows to technology companies, not teachers. Venture capital poured approximately $20 billion into edtech globally in 2021 alone, a figure that has since declined but remains substantial.

The math produces a clear comparison. It would cost $55 billion per year to make teaching a competitive profession. Instead, tens of billions flow into technology designed to make teachers unnecessary. The funds exist. The question is where they go.

Software Scales, Teachers Do Not

The economic logic driving AI into classrooms is not about learning outcomes. It is about unit economics. A software platform can serve 100 students at roughly the same cost as serving 10. A teacher cannot. This asymmetry is why investors fund edtech and why administrators adopt it.

During the COVID-19 pandemic, remote learning accelerated edtech adoption by years. K-12 technology companies raised record capital as schools rushed to put instruction online. The learning outcomes were poor, as multiple studies subsequently documented, but the market validation was powerful. School systems demonstrated that they could operate with fewer in-person teachers, even if students learned less.

The structural incentive has not changed since the pandemic ended. School budgets are under pressure. Teacher recruitment is difficult. Software offers a release valve. No venture capitalist has ever pitched a fund based on "let's pay teachers more," because that model has no margin expansion. Edtech has margin expansion built into its architecture.

This does not mean every AI education tool is harmful. Adaptive tutoring software that supplements a qualified teacher's instruction can produce measurable gains, particularly for students who need additional practice outside class hours. The distinction between supplementing and replacing is where the economics collide with pedagogy. Supplementation requires keeping the teacher and adding the technology. Replacement removes the teacher and keeps only the technology. One costs more. The other costs less. School budgets, under perpetual pressure, have a structural preference for the cheaper option.

What the Shortage Actually Costs

The teacher shortage has a measurable price, and it is not denominated in vacant positions. It is measured in student outcomes.

Students assigned to underqualified substitutes consistently score lower on standardized assessments. High teacher turnover is correlated with lower graduation rates, particularly in high-poverty schools where turnover is highest. The relationship between teacher quality and student achievement is among the most robust findings in education research.

Raj Chetty, John Friedman, and Jonah Rockoff published a landmark study in the American Economic Review in 2014 showing that a single year with a high-value-added teacher increased a student's lifetime earnings by tens of thousands of dollars. The effect compounded: students of better teachers were more likely to attend college, less likely to have teenage pregnancies, and more likely to save for retirement. Teacher quality is, by this measure, the single most influential in-school factor in a child's economic future.

The Georgetown University Center on Education and the Workforce estimates the lifetime earnings gap between a high school dropout and a graduate at roughly $400,000 to $600,000. Every student who falls through the cracks of an understaffed school system represents a fraction of that lost potential.

Scale these individual losses across millions of students in understaffed classrooms, and the aggregate cost of the teacher shortage becomes far larger than the $55 billion annual investment needed to fix it. The shortage is not just expensive for teachers. It is expensive for everyone.

The Trap

The teacher shortage and AI replacement form a feedback loop. The shortage creates political urgency. Political urgency creates openness to alternative solutions. AI companies present themselves as that alternative. If AI tools fill enough vacant classrooms, the political urgency dissipates. Legislators no longer face angry parents demanding more teachers because the seats appear filled, even if filled by screens and guides instead of certified educators.

This pattern has a precedent in higher education. American universities began hiring adjunct faculty as a temporary solution to rising enrollment in the 1970s and 1980s. Adjuncts were cheaper than tenure-track professors, required fewer benefits, and could be hired and dismissed without institutional commitment. The temporary fix became permanent. By 2021, adjunct and non-tenure-track faculty made up roughly 68 percent of all instructional staff at US colleges and universities, according to the American Association of University Professors. The cost savings that began as a stopgap became a structural feature. No one voted for it. It happened because it was cheaper.

The teacher shortage in K-12 education is following the same arc. AI is not being proposed because it teaches better than humans. It is being proposed because it costs less. Once the infrastructure is built, once school systems have invested in AI platforms, once the political crisis of empty classrooms has been managed by filling them with software, the economic and political incentives to reverse course will be minimal.

The 26.6 percent wage penalty is the number that explains the shortage. It is also the number that nobody at the White House summit mentioned. A humanoid robot walked into the building beside the First Lady. The average teacher walked out of a school that afternoon earning $18,000 less than an accountant with the same degree. The shortage is not inevitable. The data from three continents shows that. It is a choice, expressed in a budget, compounded every year it goes unaddressed.