AI in Education: What Changed in Classrooms and What Did Not

Every school leader I speak to has the same two folders. One holds the vendor emails about AI in education. The other holds the questions from staff and parents that those emails do not answer. The gap between the two folders is where most schools are standing right now: something has clearly changed, and it is not clear what.

This post is my attempt to be plain about it. What has actually changed in classrooms since artificial intelligence in education became a daily subject, what has not changed at all, and the questions a leader should ask before signing anything. No predictions, no hype.

What changed: the blank page is gone

The clearest change is at the start of a teacher's work. A lesson plan, a worksheet, a set of quiz questions, a parent email: each used to start from nothing. Now each starts from a draft produced in seconds. That is real, and teachers feel it. The hours that went into first drafts have moved to editing and checking.

Three other changes are visible in schools that have gone further than the first draft:

• Materials can match the class. A reading passage at three levels, a worksheet with larger print, an accessible version of a test: the cost of making these dropped from hours to minutes. Whether they are made depends on whether the tool knows the class.
• Every student can answer. Tools that collect an answer from all 30 students, and sort them, mean a teacher no longer has to guess from one raised hand. This is the change I think matters most and is talked about least.
• Students have a tutor at home, of some quality. Many students now ask an AI before they ask a teacher. Schools did not decide this. It happened, and the question is only how to shape it.

A Grade 9 example of the change that matters

A Grade 9 science department in a school I advise did one thing differently this year. For every lesson, the teacher asked one question that every student answered in their own words, on their own device, and the answers were sorted before the next lesson. No new scheme of work, no new textbook.

What changed for the head of department was not the lessons. It was the Monday meeting. Instead of "how did the unit go?", the question became "which objectives did students struggle with across all four classes, and what are we doing about it?" Four teachers, one shared picture of learning, and no observation needed to get it. That is AI in education doing the job it is good at: putting the learner in the room where decisions are made.

What did not change: the things that make a school a school

AI in schools has not changed what good teaching is. A clear objective, a teacher who knows who is in the room, a check for understanding before the lesson ends, and a next step that follows from it. Every one of those is still a human decision. The tools draft; a teacher decides.

It has not changed what consistency across classrooms requires either. A school where lesson quality depends on which teacher a child gets had that problem before AI and has it now, with faster drafts. A tool does not create a shared standard. Leadership does.

And it has not changed the limits of data. A tool can tell you that seven students in Grade 8 confuse two terms. It cannot tell you why, or what the teacher should do on Monday. The leaders who get this right treat AI output as a question for a professional, not an answer.

Where artificial intelligence in education goes wrong

I have seen three failure patterns, and all three are leadership problems rather than technology problems.

1. Buying before deciding. A school signs a contract, then asks what the tool is for. Six months later, usage is three teachers and a champion. Decide the job first: planning consistency, or assessment, or feedback. Then buy for that job.
2. Using AI to watch teachers. Some tools promise dashboards about teaching quality. This is the fastest way to lose a staff room. Our position is simple: we don't observe teachers. We listen to learners. The evidence a school needs is what students did and understood, not how many plans a teacher generated. The longer argument is in lesson observation tools vs learning evidence.
3. No policy, so everyone has one. Without a one-page school position, each teacher makes their own rules about student use, data, and what counts as the student's own work. A short policy, written with staff, beats a long one written for the inspector. See a school AI policy on one page.

Questions to ask before a school adopts an AI tool

• What is the one job this tool is for, and who in the school owns that job?
• Does it know our classes, or does every teacher retype the context every time?
• What does it write to a student's record, and who approves it?
• Where does student data go, and can we see and delete it?
• What does it show a school leader: evidence of learning, or activity counts about teachers?
• Can a teacher use it without the school, and does their work stay theirs?

A vendor who answers all six plainly is a vendor worth a second meeting. For what comes next, when AI tools begin to act on a teacher's behalf rather than only draft, read agentic AI in education.

How TeachBuddy AI helps

TeachBuddy AI was built around the change that matters: the learner is part of every lesson, from planning to assessment. Teachers describe each class once in a classroom profile, and every plan, worksheet, quiz and summative paper reads it. In a Space, every student answers, and the two-minute reflection after a lesson proposes updates to the class profile that the teacher approves item by item. Nothing is written silently.

For school leaders, My School gives one place where teachers submit lessons, a policy check and review checklist apply the same standard across classrooms, and a question like "who has not submitted this week" gets a plain answer. It shows learning evidence and submissions, not activity dashboards about teachers. Teachers use TeachBuddy AI free; schools add the school hub when they are ready.

Frequently asked questions

What is AI in education, in plain terms?

Software that drafts, sorts and summarises for teachers and students: lesson plans, materials, quiz questions, feedback, and answers to questions. It produces drafts and evidence. The decisions about teaching stay with the teacher.

How is AI in schools different from AI in the classroom?

AI in the classroom is what a teacher and students use in a lesson. AI in schools includes the leadership layer: policy, consistency across classrooms, data, and what evidence leaders look at. A school needs a position on both.

Should schools be worried about artificial intelligence in education?

Worried is the wrong word; deliberate is the right one. The risks are real (student data, work that is not the student's own, tools that watch teachers), and each one is managed by a decision the school makes before adopting a tool, not after.

If you lead a school and want to talk through the six questions above, contact us. If you teach, create a free TeachBuddy AI account and describe one class.