A teacher in Cleveland spent three hours last Sunday writing a quiz on the French Revolution. Her colleague across the hall asked ChatGPT and had it done in four minutes.
That story is spreading through every staff room in America right now. And most people are drawing the right alarm from it for the wrong reason.
The conclusion most people jump to: teachers are about to be fired. The conclusion the data actually supports: the core tasks of teaching are now high-exposure to AI, but the human parts of the job are what keep it from being replaced outright.
Our analysis of 500+ occupations puts high school teachers at 7/10 on AI displacement risk. The global average is 5.3. Registered nurses score 4. Electricians score 2. Teachers sit on the exposed end of the scale, not the protected end. That tracks with what people are seeing when AI writes lesson plans, quizzes, and rubrics in real time.
Here's what a 7 actually means, and what it does not.
What the AI Exposure Score Actually Measures
AI displacement risk is not about whether AI can do something related to your job. It's about whether AI can do the core value-generating tasks of your job. That distinction is exactly why teaching scores a 7 and not a 3.
Look at what a teaching week is actually made of: lesson planning, content and curriculum generation, grading and feedback, building tests and quizzes, producing differentiated materials for different reading levels, tutoring and explaining concepts one-on-one. AI in 2026 does every one of those well. Those are not side tasks. They are the bulk of the hours.
Why teachers score 7/10
The day-to-day deliverables of teaching - plans, materials, grading, explanation, test creation - are now squarely inside what AI handles well. That is high task exposure, which is the 7. What is NOT automatable is managing a room of 30 adolescents, building trust with a struggling student, and reading a class's energy and pivoting mid-lesson. Those human parts are what keep a 7 from being a 10.
Compare this to accountants and auditors, who score 8/10. The core task there, processing structured financial data against rules, is exactly what software does well. Teaching is not far behind, because the production side of the job - generating and grading content - has become just as automatable. The difference is that teaching still has a live, human core that accounting does not.
A score is task exposure, not a layoff notice. The job title tells you almost nothing. The task composition tells you everything. For teachers, the task composition has shifted hard toward what AI can do.
The Part Nobody Wants to Hear
Here's where most "AI won't replace teachers" articles stop. They declare victory, remind you that human connection matters, and send you home feeling safe.
That is the wrong takeaway. A 7 is not a safe number, and treating it like one may hurt you.
The mechanism behind a 7 is task commoditization leading to wage compression. When AI handles the deliverables that used to require professional skill and time - the plans, the materials, the grading - the perceived value of that output drops toward zero. That changes how administrators think about staffing, planning time, and pay, long before anyone talks about cutting headcount.
The wage stagnation risk
Median teacher pay sits at $64,580 with a -2% job outlook. As AI absorbs the tasks that used to justify planning time and administrative pay, the case for higher wages weakens. The job stays. The leverage disappears.
Think about the second-order effects too. A VP of Sales scores 6/10 on AI exposure, but the sales development reps below them score 8/10. When those roles collapse, the VP's entire workflow reorganizes. Teachers face the same pressure from below: instructional aides and support roles are getting absorbed, which pushes more of their work back onto the classroom teacher.
So the job does not vanish. But the ecosystem around it - the staffing ratios, the prep time allocations, the administrative burden that justified certain roles - is all shifting at once, on top of the high direct exposure of the teacher's own tasks.
A high displacement score does not mean the job disappears tomorrow. It means most of the tasks you do can now be done by AI, so the disruption is real, structural, and already underway. The human core is what you build your future on.
The Human Core That Keeps Teaching at 7, Not 10
A 7 is high, but it is not a 10. The reason teaching does not score in the danger zone is its human core. Andrej Karpathy's March 2026 analysis of 342 occupations identified three structural defenses: physical presence requirements, social trust dependencies, and real-time adaptive judgment in unpredictable environments.
High school teaching scores high on all three, and that is exactly the part AI cannot take.
-
Physical presence requirement. A classroom is a live environment. Behavior management, conflict de-escalation, the quiet kid in the third row who hasn't spoken in two weeks. None of that is remote work. None of it is automatable.
-
Social trust at scale. 81% of physicians now use AI daily, up from 38% in 2023. But patients still want a doctor in the room for hard news. Students, especially adolescents, need a human authority figure more than they need a better content delivery system.
-
Adaptive judgment under ambiguity. Every class period is a different situation. The lesson plan is a starting point. A good teacher reads the room, adjusts, improvises. That's the kind of general-purpose adaptive judgment that AI models in 2026 still fail at in embodied, social contexts.
Compare this to medical transcriptionists, who score 10/10 with a -5% job outlook. That is the danger zone: a single, well-defined, repeatable task with no social dimension and no physical presence requirement. Teaching shares the high exposure on its production tasks, but unlike transcription it still has that live, social, physical core. That gap is the difference between a 7 and a 10.
What Smart Teachers Are Doing Right Now
The teachers at risk are not the ones AI will replace. They're the ones who refuse to use AI at all, and fall behind on everything else as a result.
AI skills command a 56% salary premium across knowledge work. That number doesn't apply cleanly to public school salaries, which are constrained by collective bargaining and municipal budgets. But it applies to the adjacent roles: curriculum design, ed-tech product work, instructional coaching, corporate training. The teachers who learn to use AI tools fluently will have options the others won't.
Here's the three-part move that matters most for teachers in 2026:
Automate the administrative tail. Lesson plans, quizzes, rubrics, parent communication drafts, report card language. These are the tasks eating 2-3 hours per day for most teachers. Hand them to AI. Get that time back. Reinvest it in the work AI cannot do.
Double down on the human tasks. Mentorship. Class culture. Difficult conversations with students and families. These are the tasks that justify the role. The teachers who become known for them are the ones who get the best assignments, the department head positions, the reference letters that lead somewhere.
Build transferable AI fluency. Learn how to prompt well. Learn what AI gets wrong about your subject area. Become the person in your building who can evaluate ed-tech tools critically rather than just adopt them blindly. That is a skill worth money outside the classroom walls, even if your district never pays you for it.
Where do you stand?
500+ occupations scored 0-10. Free. Takes 60 seconds.
The Jobs Around Teaching That Are Not Safe
The classroom teacher is high-exposure but still anchored by a human core: presence, authority, and live judgment with a room full of teenagers. Some of the roles adjacent to teaching have no such anchor, and those run hotter still.
Teacher assistants score 4/10, lower than the classroom teacher, because much of the role is direct student support that is hard to automate. But the slice of that job that is pure content delivery faces rising exposure as AI handles more of it, and a budget committee can make the automation case more easily for a narrow support task than for the teacher.
Test prep tutors and standardized curriculum specialists sit in a clearly more exposed position than the classroom teacher. Tutors score 7/10 and instructional designers score 8/10, and the standardized, content-delivery end of that work is the most automatable slice of it. Adaptive AI tutoring systems now outperform human tutors on certain narrow skill-building tasks in controlled studies. Khan Academy's AI features are not the same threat to a great high school English teacher, whose classroom authority is hard to copy. They are a direct threat to the $80/hour standardized-test prep tutor whose value proposition is pure content delivery.
The second-order problem in education
The classroom teacher keeps a human core. The roles that surround and support it often do not. If your income depends on pure content delivery - standardized tutoring, instructional design, curriculum production - your exposure has no human anchor to fall back on, and your risk runs toward an 8.
The same second-order effect that hits a VP of Sales when their sales development team collapses is hitting school administrators as AI handles more of the operational load. Fewer support staff means more administrative burden pushed back onto teachers. The teacher endures, but the exposure is high and the workload is climbing.
Bottom Line
AI is already doing the production side of teaching - the planning, the materials, the grading. What it cannot do is be the human in the room. The teachers who understand that distinction will have the best decade of their careers.
Score 7/10. Above average. High task exposure, because the deliverables of teaching are exactly what AI now does well. What holds the score at 7 rather than 10 is presence, trust, and the kind of adaptive human judgment that no model in 2026 can replicate in a room full of teenagers.
High exposure is not the same as job loss. The wage compression threat is real. The adjacent role collapse is real. The teachers who treat a 7/10 score as a reason to panic, or as permission to ignore AI entirely, will both lose to colleagues who use AI on the routine work and double down on the human core.
The job endures. What you make of it is still up to you.
Find out where you stand
500+ occupations scored 0-10 on AI displacement risk. Free.