Vineeth K Kolarath
When teachers talk about AI in the classroom, one question seems to sit underneath all the others.
If AI can explain a lesson, summarise a chapter, answer questions and even prepare teaching materials, what is the teacher's role?
It is a fair question. And I think it deserves an honest answer.
Start with the job, not the tool
In my essay Automate a Process, Not a Job, I argued that a job is rarely just one task. It is a collection of different activities. Some are repetitive. Others need judgement, experience and human understanding.
Teaching is a good example.
Think about everything a teacher does in a typical week. Planning lessons. Preparing worksheets and question papers. Explaining an idea in different ways. Correcting notebooks. Noticing when a student has gone unusually quiet. Helping a child who has lost confidence. Speaking to a worried parent. Deciding what the class is ready to learn next.
And there is something else. Every day, students watch how their teachers behave, respond to mistakes and treat other people.
Some of these tasks can be supported by AI. Others are much harder to hand over.
So perhaps the useful question is not, "Will AI replace teachers?" It is this: "Which parts of teaching can AI help with, and which responsibilities must remain with the teacher?"
Where AI can genuinely help
Let us be fair to the tools. They can be useful, especially when a teacher has too much work and too little time.
An AI assistant can draft practice questions at different levels. It can suggest alternative ways to explain a difficult idea. It can simplify a paragraph, propose examples from everyday life or help prepare a quiz, a rubric or a lesson outline.
These are useful starting points.
But that is what they should remain: starting points.
A teacher still needs to check whether the questions are correct, whether the explanation makes sense, whether the material fits the syllabus and whether it is suitable for the students in that particular class.
That last part matters. AI systems can produce answers that sound convincing but are wrong. A mistake in a question or answer key can confuse an entire class.
As I explained in What Is a Thinking Machine?, a fluent answer is not necessarily a correct one.
Use AI to save time on preparation. Do not let it make the final decisions about what your students should learn.
Think of it as an assistant that can help prepare the work, but does not know the classroom in the way you do.
What stays with the teacher
This is where I believe the teacher's role matters most.
Understanding the students.
A teacher may know that one student understands a subject but becomes nervous during exams. Another may be capable but easily distracted. A third may be struggling because something difficult is happening at home.
An AI tool may have access to information about a student, if someone provides it. But information alone does not amount to understanding. A teacher has to interpret what they see, listen to the student and decide how best to respond.
Recognising real understanding.
A correct answer on paper does not always mean a student understands the subject.
Ask a student how they arrived at that answer. Ask them to explain the idea in their own words. Give them a slightly different problem and see whether they can apply what they learned.
A teacher can use these conversations, observations and assessments to build a better picture of a student's understanding. This becomes especially important when answers are so easy to obtain.
Choosing the right challenge.
In Use AI as a Mentor, Not an Answer Machine, I discussed research showing that students who relied on a standard AI assistant during mathematics practice performed worse when they later had to work without it.
Learning requires effort, but not every kind of struggle is useful. A student who is completely lost needs help. A student who is ready for a challenge needs room to think.
Deciding when to explain, when to give a hint and when to let a student try again is an important part of teaching. AI can offer suggestions, but the teacher must judge what is appropriate for the student.
Caring about the outcome.
A student who has stopped trying may need more than another explanation. They may need encouragement, patience or someone who notices that they are having a difficult day.
An AI assistant can generate encouraging words. But it does not have a personal relationship with the student or a human stake in their future.
That relationship matters.
Setting an example.
Students learn from what adults do, not just from what they say.
They notice whether a teacher admits a mistake, listens to a different opinion, treats people fairly and is willing to say, "I don't know. Let us find out."
These everyday moments help shape how students approach the world. A machine can demonstrate a response, but it cannot take the place of a responsible adult living those values.
Taking responsibility.
AI can suggest a lesson, prepare a question or offer an explanation. But someone must decide whether that material is appropriate, notice when something goes wrong and take responsibility for what happens in the classroom.
That responsibility cannot simply be handed to a tool.
None of these responsibilities is new. What is changing is the amount of routine work that AI may help teachers prepare. That gives us an opportunity to pay more attention to the parts of teaching that need human judgement.
What changes
Some familiar classroom practices will need a second look.
Homework needs rethinking.
If a student can get a complete answer from AI in seconds, a finished page of homework may tell a teacher less than it once did.
That does not mean homework has become useless. It means we should pay more attention to how the student arrived at the answer.
Ask students to show their first attempt. Ask them to explain a method in their own words. Sometimes, bring the problem back into the classroom and let students work through it where you can observe their thinking.
Assessment should make thinking visible.
A final answer is only one way to assess learning.
Ask students to explain a solution on the board, compare two different approaches, find a mistake in a worked example or apply an idea to a new situation.
These activities give teachers more opportunities to see whether students understand what they are doing, rather than simply whether they can produce an answer.
Teaching students to use AI becomes part of teaching.
Students are likely to encounter AI tools in their education and working lives. They need to learn how to use them without becoming dependent on them.
Show them how to ask for a hint instead of a complete answer. Teach them to check an AI-generated explanation against a textbook or another reliable source. Encourage them to question an answer when something does not look right.
These are useful habits beyond the classroom too.
Teachers need to demonstrate thoughtful use.
Imagine a teacher asking an AI assistant to explain a topic. The answer appears. The teacher reads it, notices a mistake and says, "This part does not look right. Let us check it."
That small moment may teach students more than a long warning about not trusting AI.
They see that using a tool does not mean accepting everything it produces.
The teacher as a guardian
There is another responsibility worth discussing: protecting students when AI enters the classroom.
Privacy comes first.
Students should not casually share personal details, photographs, marks, private conversations or information about other students with AI services. Teachers need to be equally careful when entering classroom information into these tools.
Before using a service, understand what information it collects, how that information may be used and what privacy controls are available. Schools should have clear guidance on which tools can be used and what information may be shared.
Age and suitability matter.
In September 2023, UNESCO published its Guidance for Generative AI in Education and Research.
The guidance calls for a human-centred approach to AI in education. It discusses protecting personal data, checking whether tools are suitable for educational use and considering age restrictions. It recommends that countries set a minimum age of 13 for independent conversations with generative AI platforms.
That is a recommendation in UNESCO's guidance, not a universal legal rule for every classroom. Schools must consider the rules that apply to them, the terms of the services they use and the age and needs of their students.
The broader point is simple: a tool should be checked before it is introduced to children. Being available does not automatically make it suitable.
Fairness matters too.
Not every student has a computer, a personal phone or a reliable internet connection at home.
If assignments begin to assume that every child can use an AI service, some students may be placed at a disadvantage through no fault of their own.
Teachers need to be aware of these differences. Classroom learning should not depend on resources that some students cannot access.
Start small
If you are a teacher who has not used AI much, you do not need to change your entire teaching method.
In Why Your First Project Should Be Small, I suggested that beginners in engineering start with one small thing that works.
The same advice applies here.
Choose one task. Perhaps you want to prepare practice questions for a chapter. Try using AI to draft them. Check every question, correct what needs correcting and see whether it actually saves you time.
Then try one small change in the classroom. Ask students to show their first attempt before they use an AI tool. Listen to how they explain their thinking. Notice what you learn about their understanding.
Do not begin with the assumption that AI will improve everything. Try it, observe what happens and decide what is worth keeping.
A few careful experiments will tell you more than a large plan built on assumptions.
The role becomes more important, not less
It is natural for teachers to feel uneasy about AI in the classroom. Some of that unease is useful. It reminds us to ask questions before adopting a new tool.
But I do not think the arrival of AI makes teachers less important. It makes us look more carefully at what good teaching actually involves.
AI may help prepare a worksheet, suggest an explanation or draft a lesson. Those things can save time.
But a student still needs someone who notices when they are struggling, recognises when they have understood an idea, chooses the right challenge and helps them find the confidence to keep going.
That is the work of teaching.
The tools may change. The responsibility to help a student learn remains.
Try this
If you are a teacher, try a small exercise this week.
Choose one lesson you will teach soon. On a sheet of paper, list the tasks involved: preparing the lesson, explaining the topic, giving students practice, checking their work and following up.
Next to each task, write one of three letters:
- A — AI could help you prepare or draft this.
- T — This needs your judgement and attention as the teacher.
- S — This is the student's own work and should not be done for them.
Look at the list. Choose one A task and try an AI tool. Check the result carefully.
Then protect the T and S tasks. Do not hand over your responsibility as a teacher, or your students' opportunity to think and learn.
In Bring a Rough Idea, I asked readers to begin with a rough idea.
For teachers, this page can be a rough map: where a machine may help, where a student must do the work, and where the teacher's judgement matters most.