The Message Beneath the Conversation

In 2025, Microsoft announced a $4 billion investment in AI education. The same year, the latest national surveys showed that more than half of America's K–12 teachers were using AI tools in their classrooms — and that the policy guidance, training, and academic-integrity infrastructure that should have arrived with those tools had not.

That gap — between what is being purchased for schools and what teachers and students are actually equipped to do with it — is the story of American education in 2026. It is also the story of American education in 2016. And 2006. And 1996.

The crisis we keep talking about is not new. The framing we keep using to talk about it is wrong. And the conversation we are now having about artificial intelligence in schools is happening on top of a foundation that has been quietly cracking for forty years.

This paper is about that foundation. About what the data actually shows. About who benefits when we keep getting the diagnosis wrong. About what the system has been trying to tell us for forty years that we have not been willing to hear. And about why no amount of technology will save us from a system we refuse to examine honestly. Because the most important conversation in K–12 right now is not about AI. It is about the conditions underneath it.

If you searched American newspapers for the phrase "teacher shortage," you would find it in 1982. And 1992. And 2002.

And then every year since 2002. Not another decade. Every year.

The headline has been written, retired, and re-written so many times that it has become a kind of seasonal weather report — something everyone expects, no one is surprised by, and no one fixes.

In 1984, Linda Darling-Hammond published a report for the RAND Corporation called The Coming Crisis in Teaching. She was already, in 1984, describing the cycle as decades old. The report documented what would become a pattern: shortages emerge, federal programs respond, conditions briefly improve, the programs are cut, and the shortage returns. We have been on that loop for fifty years.

Recent research from Matthew Kraft and Melissa Arnold Lyon (2024) tracked the health of the teaching profession across half a century using four independent measures: how prestigious teaching is considered, how interested college students are in entering it, how satisfied current teachers are, and how many students enroll in teacher preparation programs. The four lines tell the same story. A sharp decline in the 1970s. A real recovery in the 1980s and 1990s. Stability through the early 2000s. And a sustained, uncorrected decline that began around 2010 and has worsened since.

The most consequential reframing of this conversation came from Richard Ingersoll at the University of Pennsylvania. In 2001, he published a paper titled Teacher Turnover and Teacher Shortages: An Organizational Analysis. He looked at the federal data from the National Center for Education Statistics and reached a conclusion that should have changed everything we did next:

Between 1987 and 2021, the number of students in American schools grew by 22 percent. The number of teachers grew by 61 percent — nearly three times as fast. We have been graduating teachers at an accelerating pace for four decades. And every year, roughly 15 percent of them leave their jobs. Most do not leave to retire. Most leave because of the conditions inside the schools where they work.

By June 2025, approximately 411,500 teaching positions in American public schools were either vacant or filled by teachers not fully certified for their assignments — about 1 in 8 positions nationally (Learning Policy Institute, 2025). The shortage is real. The framing of why we have it has been wrong for fifty years.

There is no supply problem in the aggregate. There is a retention problem in specific schools, in specific subject areas, and under specific organizational conditions — and those conditions are known. They have been documented across two decades of research from Ingersoll, the Wallace Foundation, Bryk and Schneider, Fullan, and others. We are not waiting on better data. We are not waiting on a federal solution. The levers that produce teacher retention and stronger student outcomes are within district authority right now, today — without a redraft of the entire system. We have known what to do. We have not done it.

Sources: Ingersoll (2001, 2023); Kraft & Lyon (2024); Learning Policy Institute (2025); Nguyen et al. (2024); NCES SASS/NTPS surveys.

If we have known since the early 2000s that retention is the operative variable, why does the conversation continue to be framed as a shortage?

Because the framing determines the response. And the two responses are not the same.

If we call it a shortage	If we call it a retention problem
The problem is supply.	The problem is conditions inside schools.
Solution: recruit more teachers.	Solution: change what teachers experience after they walk in the door.
Action: signing bonuses, fast-track certification, more teacher preparation programs.	Action: protected time, coherent curriculum, present and supported leadership.
What it costs: money.	What it costs: changed adult practice.
Politically convenient. Technically simple.	Politically uncomfortable. Adaptively hard.
Has been the default response for fifty years.	Has been the documented answer for twenty-five.

This is the difference between a problem you can solve by adding more, and a problem you can only solve by changing what you do. American education has spent forty years adding more — more teachers, more programs, more requirements, more reforms — and almost nothing changing what we do. The visible cost is a recurring shortage. The invisible cost is that the schools where conditions are worst are the same schools where the shortage is most severe — the schools serving the children who already have the least.

If you ask most parents, board members, or community members what teachers do during their "prep period," you are likely to get a vague answer. Grading, maybe. Lesson planning. The administrative work that goes with running a classroom. It sounds like one thing. It is treated like one thing. But it is two things, and conflating them is one of the quietest ways we have undermined American teaching.

Teachers need two distinct kinds of protected time. Both are essential. Both are routinely lost. And neither one can substitute for the other.

Individual planning time

This is the time a teacher needs alone — to think, to design, to prepare. It is when she analyzes what students did or did not understand yesterday and decides what tomorrow needs to look like. It is when she reads through a unit and decides which standard to teach explicitly and which to weave in. It is when she designs a formative assessment that will actually tell her what students know. It is when she prepares the materials, the slides, the questions, the differentiated supports. It is the cognitive heart of teaching.

This work is solitary by nature. It requires sustained, uninterrupted blocks of focused time. It cannot be done in five-minute increments between bus duty and a parent phone call. It cannot be done while supervising lunch. When this time is taken away, instruction becomes improvisational. The teacher walks into class and does her best, but what she has had time to actually plan is a fraction of what she is being asked to deliver.

Professional Learning Community time

This is the time teachers need with each other. It is when a fourth-grade team sits down and looks at student writing samples together and decides what "proficient" actually means in their school. It is when a math department compares the ways three different teachers taught the same concept and figures out what worked best. It is when teachers calibrate their expectations so that students moving from one classroom to the next experience a coherent education — not three different versions of the same grade level.

PLC time is what turns a school full of individual teachers into a coherent system. Without it, every classroom becomes its own island. Students learn whatever their assigned teacher happens to emphasize, in whatever way that teacher happens to teach it, with no shared standard for what success looks like. This is how we end up with classrooms in the same building delivering radically different educations under the same school name.

What happens when we collapse them

Most American school districts give teachers a single block of time and call it "prep period." Then they fill it with mandated meetings, coverage when colleagues are absent, parent phone calls, IEP paperwork, and the fifty other administrative obligations that have been added to the teaching job over the last twenty years.

The teacher ends up with neither. Not enough alone time to plan. Not enough collaborative time to align with her team. The two functions — individual instructional design and collective coherence — collapse into a single overcrowded slot that serves neither one. And then we wonder why instruction is uneven, why standards are inconsistently implemented, why teachers feel they are constantly behind, and why the work cannot be sustained.

Protecting both kinds of time is not a perk. It is not a luxury. It is the structural minimum required for instruction to be coherent. And it is one of the conditions we already know how to fix — today, without legislation, without new funding, by simply protecting the master schedule from being eroded by everything else the system has added on top of it.

If the answer has been documented for two decades, and the conditions are within district authority, why has the system not corrected itself?

Because the people and institutions with the most influence over American education have, on the whole, no incentive to fix it. The professional development industry sells workshops the field has known for decades produce only 5 to 10 percent transfer to classroom practice without coaching (Joyce & Showers, 2002). The curriculum publishing market depends on districts adopting new programs every two to three years — even though research is unambiguous that initiative churn produces no sustained implementation. The educational technology sector, projected to exceed $400 billion globally, depends on continuous adoption cycles, recurring license fees, and the constant onboarding that high teacher turnover guarantees. And every district's operating budget is, quietly, easier to balance when high-cost veterans are replaced by lower-cost novices each year.

None of these actors is acting in bad faith. The professional development provider believes the workshop helps. The publisher believes the new program is better than the last one. The vendor believes the tool will give teachers time back. What they share is an absence of incentive to address the foundation underneath all of it. The system is not corrupt. It is incentive-aligned to its own perpetuation. And that is harder to change than corruption would be.

This is where leaders have more agency than the political economy of American education would suggest. Because the same dysfunction that the system rewards is also the system's clearest communication.

Severe teacher shortage in your buildings? That is a signal. Stressed, demoralized, exhausted staff? That is a signal. A sense that this year's initiative is being layered on top of last year's unfinished work? That is a signal. A district being asked to integrate AI into classrooms whose curriculum is not coherent and whose teachers have not been trained? That is a signal too. What looks like dysfunction is actually information. The system is telling you something. Read it.

These signals are not problems to be managed away. They are messages telling district and site leaders that the foundation is not stable enough to support what is being built on top of it. And the right response, when those signals are this loud, is not to push harder. It is to pause.

This is not new advice. Mike Schmoker, in Focus (2nd ed., 2018), instructs schools to "declare a moratorium on new initiatives" so that the essentials can take hold. The broader research converges: districts with two to five priorities outperform those carrying extensive lists, with Reeves identifying six as the practical ceiling and Fullan and Stiegelbauer documenting initiative overload as far back as 1991. We have known about this problem for thirty-five years.

A moratorium on new initiatives is not an admission of failure. It is an act of leadership. It says: before we add the next thing, we are going to finish the last three. We are going to listen to the staff who have been telling us, in every survey and every exit interview, what they need. We are going to communicate with transparency about why we are making the decisions we are making and how teachers' professional expertise will be honored in the work ahead. We are going to check the egos that have made it easier to launch than to listen. We are going to remind ourselves of the professional responsibility we share to the children in our buildings, and reignite the why that brought us into this work — protecting it in our teachers and our students. And we are going to build coherence together, in the time we already have, with the people we already have.

This is the work no vendor can sell. No technology can replace it. No grant can fund it. It is the work of leadership choosing the harder, slower, deeper thing — and trusting that the people in the buildings, when given the conditions to do their work well, will do it.

The case for AI literacy in K–12 is real. Students will enter a workforce, civic life, and information environment in which AI fluency is foundational. Districts that fail to prepare them are not protecting students from AI — they are exposing students to it without scaffolding. The frameworks for thoughtful integration exist, and the equity stakes are urgent.

But here is the part most of the field is not saying out loud. AI deployed on top of a coherent instructional system extends a teacher's capacity to deliver excellent instruction. AI deployed on top of an incoherent system does not resolve the incoherence. It absorbs it. And eventually it amplifies it.

The data shows the early shape of this. By the 2024–2025 school year, 53 percent of teachers reported using AI for school work — a jump of more than fifteen percentage points in a single year (RAND, 2025). Yet over 80 percent of students reported that no teacher had explicitly taught them how to use AI for schoolwork. Only 34 percent of teachers reported having school or district policies on AI-related academic integrity. Only 45 percent of principals reported their school or district had any AI policies at all.

And the equity gap in training is severe. By fall 2024, 67 percent of low-poverty districts had provided AI training for teachers, compared with 39 percent of high-poverty districts (RAND, 2025). Adoption is racing ahead of the policy and training infrastructure that would make adoption work. The teachers and students least likely to receive guidance are the ones already absorbing the consequences.

Teachers are reaching for AI for the same reason they have always reached for whatever tools are available: because the system has not given them what they actually need. The reality on the ground is more complicated than the policy debates suggest. Teachers are navigating a dozen or more technology platforms that often do not work together, ensuring those tools function reliably enough to teach with, and trying to keep students engaged while doing it. AI tools are easy to drop in and often more engaging than what the curriculum provided. But a fun AI activity is not the same thing as instruction aligned to essential standards. With clarity about what students must learn and a guaranteed and viable curriculum to anchor the work, supplemental AI tools can extend a teacher's capacity. Without those anchors, AI becomes another fragmenting force in an already fragmented day.

And the whiplash compounds the problem. AI helps with one task; then a new district policy restricts it for safety reasons or out of uncertainty about what is permissible. Teachers change their practice — again. A tool changes its pricing, gets blocked, or stops working the way it did. Teachers pay for an alternative themselves, only to have that one blocked too. They are absorbing the cost of a system that cannot decide what it wants to be, while still being expected to deliver coherent instruction every day. The appearance of innovative integration is being mistaken for the harder work the system is still avoiding.

And the data tells only part of the story. Teachers are also navigating the human cost of unsupported AI deployment: students misusing tools to bypass learning, academic integrity violations they have no clear policy to address, and the added workload of detecting AI-generated work — all without training, without institutional support, and often without clear answers from the district on what counts as appropriate use. The technology arrived faster than the guidance. Teachers absorbed that gap.

This is not an argument against AI in schools. It is an argument about sequence. The districts that will produce the strongest AI implementations over the next five years will not be the ones moving fastest. They will be the ones whose foundations are coherent enough to make AI integration meaningful in the first place.

None of these moves are new. None require federal action. None require a vendor partnership. All of them are within district authority right now.

A guaranteed and viable curriculum. Without one, AI generates lesson plans for objectives no one has actually agreed on. With one, AI helps teachers deliver a coherent curriculum more effectively.

And to be clear: this is not a one-size-fits-all mandate. Curriculum coherence is not the same as curricular rigidity. A guaranteed and viable curriculum provides the baseline — the shared agreements about what students will learn and when — that allows teachers to exercise professional judgment within a coherent structure rather than improvising in isolation. Technology can make some teachers' work easier and more effective. But not all teachers, and not all technology, and not without the conditions that allow discernment about when a tool serves learning and when it substitutes for it. There will never be a one-size-fits-all approach for instruction — just as there will never be one for our students. The structure is what makes professional judgment possible.

Protected planning and PLC time. Both kinds. Separately. Without these, AI becomes a substitute for the conditions that produce strong instruction in the first place.

Instructional leadership in every building — visible and known throughout the building. Not present in name only on an org chart. Visible in classrooms, known by teachers, present in the daily work. No tool, no platform, and no dashboard substitutes for the leader who knows what is happening in their school.

Comprehensive induction and mentoring for novice teachers. Ingersoll and Strong's 2011 meta-analysis found that comprehensive induction — mentoring paired with common planning, reduced teaching load, and collaborative networking — reduces first-year teacher turnover by more than 50 percent. The teachers we spend the most public dollars to recruit are the ones leaving fastest. The intervention with the largest documented retention effect for that population is structural, district-controlled, and within reach this year.

A moratorium on new initiatives until the current ones are working. Adding AI to a district that has not finished its last three initiatives is not innovation. It is incoherence with a more expensive vendor contract.

Working conditions that retain experienced teachers. Retention by itself does not guarantee strong outcomes — some districts retain through compensation and contract protections rather than through coherent practice. But losing veteran teachers means losing the institutional knowledge that makes any tool, any reform, and any improvement effort actually work in practice. That loss is one of the most expensive a school can sustain, and the system creates it every year.

This paper was written for educational leaders. But the conditions it describes — coherent curriculum, protected teacher time, present leadership, manageable initiative load, retention of experienced teachers — will not change without political will. In American public education, political will comes from the people who show up to school board meetings, ask questions, and hold elected officials accountable. If you are a parent, a guardian, a grandparent, a community member, or anyone who cares about the children in your local schools, you have more authority than you probably realize.

Your tax dollars fund this system. Your children sit inside it. The conditions described in this paper are decisions, not destiny — and they are being made every month, in public, by people who answer to you. The choice your community is making, whether it knows it or not, is between using its influence to push for the conditions that actually help students thrive in their schools — or accepting another decade of endless initiatives and personal political agendas while the foundation continues to crack.

What follows is not a list of questions for board meetings. Spotlight meetings and district marketing already produce polished answers to questions. What follows is a set of look-fors — observable evidence in documents, schedules, hallways, and behavior that surfaces what is actually happening in your district. They do not require insider knowledge to interpret. They require attention. And as you go through them, compare across schools within your district. Many of these will read differently at the highest-need school than at the flagship — and that gap is itself one of the most important signals available.

If you cannot find this evidence, or if the answers turn vague when you ask — that itself is information. The system is signaling. You can decide what to do with that signal.

These look-fors are not about making trouble. They are about empowering professional responsibility inside your school system — giving the educators in your community permission, and protection, to do what the research says works. Carry the evidence with you. Cite it at board meetings, in emails to leadership, in every room where decisions are made. Educators inside the system often cannot speak the most uncomfortable truths in public. Community members can. That is not adversarial. It is what democratic accountability of a public institution looks like.

You do not have to be an expert in instructional systems to look. You only have to be willing to look honestly, and keep looking until the picture becomes clear. Empowering families to participate in this work is not a soft skill. It is a structural lever. The community whose voice helped define what learning looks like in their district is the same community that defends that work when opposition organizes.

Throughout this paper, the focus has been on conditions, structures, and signals. But behind every signal is a person. Behind every data point is a teacher who showed up this morning despite everything the system has asked her to absorb, a principal still trying to lead inside an incentive structure that punishes the long work, a paraprofessional, a counselor, a bus driver, a custodian. Every one of them is still standing in this profession by choice. They deserve to be seen as more than variables in someone else's analysis.

We owe them, at minimum, two things.

The first is perspective. Generational understandings of respect, professionalism, and authority have shifted dramatically in the last thirty years. The teacher who entered the profession in 1995 and the student sitting in her classroom in 2026 are operating from genuinely different cultural frames about what respect looks like, what civility sounds like, what professionalism feels like in practice. Most of the daily friction inside our schools — between staff and students, between staff and families, between staff and administrators — is rooted here. Not in bad intentions. In the absence of shared perspective. Perspective is something a highly individualistic society loses easily: every actor has a story that, if heard, would make their behavior more legible. Every leader has a constraint the people around them cannot see. Every teacher reaching the end of her capacity has reasons the data does not capture. Every student behaving badly is communicating something the adults have been trained to manage rather than to understand. None of this excuses harm. But naming it is the first step toward addressing it without making the problem worse.

The second is honesty about what the data actually shows. In recent years, many districts have reduced their reported suspension rates and celebrated that reduction publicly — while educators inside those buildings have watched on-site violence and disrespect increase. Reporting patterns are not behavioral reality. A district that changes how it codes incidents has not changed what is happening in its classrooms. It has only changed what shows up in its data. Educators in the buildings know the difference. They are the ones absorbing it. And presenting them with data that contradicts what they see every day is one of the fastest ways to lose their trust — and a quietly common practice across American school leadership.

The students in classrooms today will inherit the world we are building for them. That world includes AI — whether we like it or not — and they deserve to be prepared for it.

But preparation is not the only thing they deserve. They deserve to be taught by teachers who are not absorbing the cost of an incoherent system in their daily work. They deserve curriculum the adults around them have agreed on. They deserve leaders who are visible and known in their buildings. They deserve schools whose working conditions are worthy of the work they are asking children to do.

They also deserve an honest accounting of what forty years of deferred action has cost them. Global IQ scores — which rose steadily for most of the 20th century in what researchers called the Flynn Effect — have been declining in industrialized countries since approximately 2000 (Bratsberg & Rogeberg, 2018). The British Cohort Study (2024) documents a cumulative decline of up to five IQ points in verbal skills and structured reasoning among post-1970 cohorts. German secondary school students tested in 2020 after six months of pandemic-affected schooling showed substantially lower intelligence test scores than comparable cohorts tested in 2002 and 2012, with no signs of catching up after another full school year. International reading scores declined by 33 percent of a standard deviation between 2016 and 2021 — equivalent to more than a year of schooling. United States fourth-graders' math scores declined 18 points from 2019 to 2023; eighth-graders declined 27 points (TIMSS, 2023). Creativity scores among American students have been declining since the 1990s.

This is not a genetic shift. Researchers attribute the decline to environmental, social, and especially digital factors — the conditions we create for children, not the children themselves. And critically, the research suggests the decline is reversible if targeted efforts are made to address the academic deficits that have occurred. We have the knowledge. We have the levers. We are choosing not to use them.

And both students and teachers deserve to keep the spark that brought them through the schoolhouse door in the first place — the curiosity, the inspiration, the sense that this work matters. We are smothering that spark with fractured initiatives that leave the people in our buildings fatigued, frustrated, and angry. The moral case is not only about test scores or retention rates. It is about whether the people inside our schools — children and adults alike — get to keep what makes the work worth doing.

The teacher shortage is not the foundation problem. It is what the foundation problem looks like from the outside. The deeper truth is harder to face.

We have known the answer for forty years.
We have not done the work.

That gap is not a knowledge gap. It is not a resource gap. It is not the failure of any single leader, any single teacher, or any single district. It is something larger and harder than any of those things.

It is a system gap.

Now you know. There is no longer a way to say you don't.

We have known what produces teacher retention since 2001. We have known what produces durable instructional improvement since the Bryk and Schneider Chicago studies of the early 2000s. We have known what produces effective school leadership since the Wallace Foundation's research on principal pipelines. The evidence has been here. The will to act on it has not.

That gap is not abstract. It costs students every day it remains open. It costs teachers their careers, their health, and their faith in the profession. It costs communities the public schools they were promised. And no amount of technology, no amount of innovation, no amount of investment in the latest tool will close it until we are willing to look honestly at the system itself — and do the work we have already known how to do for forty years.

Now that we know better, it's our job to do better.