Everyone Is Using AI, but Schools Are Behind on Training, Policy, and Listening

The AI problem in schools is not just adoption. It is experience.

Artificial intelligence is already inside education. Teachers are using it to draft lesson plans, build quizzes, simplify reading passages, write emails, translate content, generate examples, and reduce administrative workload. Students are using it to brainstorm, summarize, study, write, code, translate, and sometimes bypass assignments.

The issue is no longer whether AI will enter schools. It already has.

The deeper problem is that many schools are still making AI decisions without a full understanding of what teachers, students, parents, advisors, accessibility staff, librarians, IT teams, and support staff are experiencing every day.

That creates a dangerous gap.

On one side, AI is becoming part of everyday academic work. On the other side, formal training and policy are still thin. A May 2026 Gallup/Walton Family Foundation study found that only 18% of U.S. public K–12 teachers said they had received formal guidance from school administrators on how AI tools should be used. Across ten AI-related work tasks, 34% of teachers said they had received no guidance at all, while 48% said they had only informal guidance.

That means the AI future is not being planned only in boardrooms or technology committees. It is already being improvised in classrooms.

Teachers are not waiting for policy because the work cannot wait

Teachers are using AI because the workload problem is real.

Lesson planning takes time. Differentiating materials takes time. Writing parent messages takes time. Creating rubrics takes time. Building examples, discussion prompts, quizzes, and feedback takes time. Teachers are not using AI only because it is trendy. Many are using it because it offers some relief from an overloaded system.

The same pattern is visible outside the United States. A 2026 National Education Union survey in England found that 76% of teacher respondents were using AI tools for day-to-day work, up from 53% the previous year. The most common uses were resource creation, lesson planning, and administrative tasks. But the same survey found that 49% of schools had no AI policy for staff or students.

This is the contradiction at the center of AI in education: teachers are already adapting, but institutions are not always listening, training, or governing at the same speed.

When schools do not provide formal guidance, teachers still have to make decisions. Should they use AI for lesson plans? Can they upload student writing? Can they use AI to generate feedback? Should students be allowed to use AI for outlines? What counts as cheating? What counts as support? What counts as learning?

Without a shared framework, every classroom becomes its own AI policy lab.

The missing piece is teacher experience

Most AI conversations in education begin with tools, risk, cheating, or efficiency. Those are important, but they are not enough.

The first question should be:

What are educators actually experiencing?

A teacher may not describe the AI issue in technical terms. They may say:

• “I do not know what I am allowed to do.”
• “Students are using AI, but I cannot prove it.”
• “I want to use AI to save time, but I am worried about student privacy.”
• “The district gave us a tool, but not enough training.”
• “The policy sounds good, but it does not match classroom reality.”
• “I need examples, not slogans.”
• “I do not want AI to replace the human part of teaching.”

These statements matter because they show how policy is felt at the classroom level.

If policy is written without teacher experience, it becomes abstract. If training is designed without teacher input, it becomes generic. If AI tools are purchased without classroom testing, they may solve the wrong problem.

UNESCO’s AI competency framework for teachers makes this point indirectly: AI changes the traditional teacher-student relationship into a teacher-AI-student dynamic, which requires rethinking teachers’ roles, competencies, human agency, ethics, pedagogy, and professional learning. UNESCO also notes that few countries have clearly defined these competencies or built national teacher-training programs around them.

That means teacher voice is not optional. It is part of the system design.

Student voice is also missing

Students are often discussed as the problem: cheating, shortcuts, plagiarism, distraction, dependency.

But students are also experiencing confusion.

Some teachers allow AI for brainstorming. Others ban it. Some classes require disclosure. Others never mention it. Some students use AI as a tutor. Others use it to avoid reading. Some students have paid tools. Others rely only on free versions. Some students know how to verify AI output. Others assume the answer is correct because it sounds confident.

From the student perspective, the rules are often inconsistent.

This matters because AI literacy is becoming part of academic literacy. Students need to know when AI is useful, when it is unreliable, when it must be disclosed, and when it damages learning.

A policy that only says “do not cheat with AI” is too weak. Students need a more practical language:

• AI for support
• AI for practice
• AI for brainstorming
• AI for feedback
• AI for production
• AI as prohibited substitution

The central distinction should be simple:

Does AI support the student’s learning, or does it replace the student’s learning?

Students should be part of that conversation. Not because they should write the whole policy, but because they know how AI is actually being used outside adult supervision.

Parent experience matters because AI changes trust

Parents are also being pulled into the AI transition, often without clear information.

They may hear that AI is helping teachers save time. They may also hear that students are cheating. They may see children using AI for homework and not know whether it is allowed. They may worry about privacy, screen time, bias, or whether AI tutoring is replacing human support.

Parents do not need technical AI jargon. They need clear answers:

• Is my child allowed to use AI?
• Is the school teaching responsible AI use?
• Are teachers trained?
• Is student data protected?
• Are AI tools being used for grading or feedback?
• Will AI reduce human support?
• What happens if my child uses AI incorrectly?
• How will students learn real writing, reading, math, and reasoning skills?

If parents are not included, AI policy can look like another top-down technology shift. That damages trust.

The public debate around AI tutoring shows why this matters. Reporting on the NEU survey noted teacher concerns that AI could weaken critical thinking and reduce human interaction, especially for students who need more than academic support. Whether one agrees fully or not, the concern is legitimate: education is not only content delivery. It is relationship, feedback, motivation, identity, and belonging.

Support staff and specialists need a seat at the table

AI policy cannot be written only by administrators, IT, and curriculum leaders.

Schools should also listen to:

• Special education staff
• Accessibility coordinators
• English language learner specialists
• Librarians
• Academic integrity officers
• Counselors
• Advisors
• Instructional designers
• Teaching assistants
• Data privacy officers
• Front-desk and student support staff
• Technology support teams

These groups see different failure points.

Accessibility staff may see AI tools that are not usable by all students. Librarians may see weak citation practices and fake sources. Counselors may see students over-relying on AI instead of seeking human help. IT teams may see privacy and account risks. Instructional designers may see assignments that no longer measure the intended learning outcome. Teaching assistants may see inconsistent grading and unclear student expectations.

A good AI policy must be built from these operational realities.

The problem with top-down AI policy

Top-down policy is fast. It is also fragile.

A district, college, or university can quickly publish a statement saying AI must be used ethically, responsibly, transparently, and in alignment with academic integrity. That sounds good, but it may not help a teacher on Monday morning.

The practical questions remain:

• What does responsible use mean in a fifth-grade writing class?
• What does AI disclosure look like in a biology lab?
• Can AI be used to translate parent communication?
• Can a student use AI to understand a reading passage?
• Can an instructor use AI to draft feedback?
• Can a teacher paste student work into an AI tool?
• Can AI-generated quiz questions be used without review?
• Who decides which tools are approved?
• What happens when the AI tool gives wrong information?

If policy does not answer classroom-level questions, it becomes symbolic.

The goal should not be to produce the longest AI policy. The goal should be to produce a policy that matches lived experience.

AI policy should be built like user experience research

Schools already use surveys, focus groups, interviews, pilots, and feedback loops for many programs. AI policy should be built the same way.

Before writing rules, institutions should ask what different groups are experiencing.

Teachers can describe workload, assessment concerns, lesson planning needs, student behavior, and classroom ambiguity.

Students can describe how they use AI, where rules are unclear, and what kinds of support would help them learn instead of shortcut.

Parents can describe trust concerns, privacy worries, and what communication they need from schools.

Support staff can describe operational risks, accessibility gaps, and student support issues.

Administrators can describe governance, procurement, liability, and implementation constraints.

IT and data privacy teams can describe tool risk, account security, vendor agreements, and data handling.

The point is not to let every group veto policy. The point is to prevent policy blindness.

AI will affect different people differently. A student with a disability may experience AI differently from a student using it to write essays. A teacher with 150 students may experience AI differently from an administrator reviewing procurement. A parent of a younger child may have different concerns than a college student preparing for the workforce.

Good policy must see the whole system.

The equity issue is not just access to AI tools

Many AI equity conversations focus on who has access to paid tools. That matters, but it is not the whole issue.

The larger equity question is:

Who gets trained, who gets heard, and who gets protected?

If only some teachers receive AI training, students receive uneven instruction. If only some students understand AI disclosure, academic integrity enforcement becomes uneven. If only some parents know the policy, family support becomes uneven. If only some schools have approved tools, AI use becomes tied to local funding and informal experimentation.

Gallup found that teachers in higher-income schools were somewhat more likely than teachers in higher-need schools to receive some type of AI guidance, especially around preparing to teach and modifying student materials.

That matters. AI could reduce workload and improve access, but without policy and training, it could also widen gaps.

An under-resourced school may end up with unclear rules, free consumer tools, weak privacy protections, and no training. A better-resourced school may have approved tools, training sessions, AI literacy lessons, and administrator support.

That is not just a technology divide. It is an institutional readiness divide.

Experience should shape the policy categories

A practical AI policy should emerge from real school experience and then classify AI use into understandable categories.

For example:

Acceptable support

AI can help with brainstorming, study questions, grammar support, translation assistance, or practice explanations when the assignment permits it.

Disclosed assistance

AI can help create outlines, revise drafts, generate examples, or support coding practice when the student or teacher clearly discloses how it was used.

Human-reviewed professional use

Teachers can use AI to draft lesson materials, rubrics, or parent communication, but the teacher remains responsible for accuracy, tone, privacy, and instructional quality.

Restricted or prohibited use

AI should not be used to submit work as original, fabricate citations, bypass required learning, make high-stakes decisions, expose private student data, or replace legally required services.

These categories are better than a generic allow/ban framework because they reflect actual experience.

Training should start with stories, not software

Most AI training begins with tools: “Here is how to use ChatGPT,” “Here is how to prompt,” or “Here is our approved platform.”

That is useful, but it should not be the starting point.

Training should begin with real scenarios:

• A student submits a polished essay far above their normal writing level.
• A teacher wants to use AI to create feedback comments for 120 essays.
• A parent asks whether AI is grading their child’s work.
• A student with dyslexia uses AI to simplify a reading passage.
• A teacher uses AI to translate a message to a family.
• A student uses AI to generate fake citations.
• A school buys an AI tutoring tool without teacher input.
• A student relies on AI so heavily that their writing development stalls.

These are experience-based cases. They help educators discuss judgment, not just tool operation.

AI literacy is not only knowing how to prompt. It is knowing how to decide.

AI governance must preserve human agency

The strongest AI policies will not be the ones that maximize automation. They will be the ones that preserve human agency.

UNESCO’s teacher framework explicitly emphasizes human-centered thinking, ethics, pedagogy, AI foundations, and professional learning. That framing is important because schools are not factories for content production. They are human learning environments.

AI can support education, but it should not quietly redefine education around convenience.

If teachers use AI to save time, the saved time should strengthen teaching, not justify more workload. If students use AI for support, the support should build skill, not hide skill gaps. If administrators use AI to scale services, the scaling should not erase relationships. If vendors sell AI tools, schools should evaluate them against learning goals, not marketing claims.

The essential question is not “Can AI do this?”

The question is:

Should AI do this in this educational context, for this learner, with these risks, under these rules?

That question requires human judgment.

The future of AI in education should be co-designed

AI policy should not be something done to educators. It should be something built with educators.

It should not be something announced to students. It should be something explained to students.

It should not be something hidden from parents. It should be something communicated clearly to families.

It should not be something owned only by IT. It should be shared across teaching, learning, privacy, accessibility, assessment, student support, and leadership.

The schools that handle AI well will not necessarily be the schools with the most advanced tools. They will be the schools that build the best listening systems.

That means:

• Listening before buying
• Listening before banning
• Listening before mandating
• Listening before training
• Listening before assuming all AI use is either innovation or cheating

AI is changing education, but education should not be redesigned around AI alone. It should be redesigned around the people who experience learning every day.

Final analysis

The AI gap in education is not only a technology gap. It is a listening gap.

Teachers are using AI, but many lack formal guidance. Students are using AI, but often face inconsistent expectations. Parents are concerned, but often lack clear communication. Support staff see risks, but are not always included. Administrators are pressured to act, but may not have enough ground-level evidence.

This is how institutions make weak technology decisions: they respond to hype, fear, and vendor pressure before they fully understand lived experience.

The better path is slower, but stronger.

Schools need AI policy and training, but they also need voice architecture: structured ways for educators, students, parents, and staff to shape how AI is used.

Because the real question is not whether everyone is using AI.

They are.

The real question is whether everyone affected by AI gets heard before the rules are written.

References

1. Gallup / Walton Family Foundation — Most Teachers Receive No Formal Guidance on AI Use. Published May 27, 2026.
2. National Education Union — State of Education: AI. Published April 2, 2026.
3. UNESCO — AI Competency Framework for Teachers. Published August 8, 2024.
4. The Guardian — Pupils in England are losing their thinking skills because of AI, survey suggests. Published April 2, 2026.