AI in Education
How Can AI Be Used in Schools? 7 Practical Uses That Already Work
A practical guide to using AI in schools: teacher lesson planning and question papers, first-pass grading of written work, personalised practice for weak students, attendance and admin automation, early warning for dropouts, and accessibility for students with disabilities. Includes what schools should not use AI for, a rollout order that works, and the Indian policy context including CBSE's AI subject and NCERT's DIKSHA content.
Admin ·
A practical guide to using AI in schools: teacher lesson planning and question papers, first-pass grading of written work, personalised practice for weak students, attendance and admin automation, early warning for dropouts, and accessibility for students with disabilities. Includes what schools should not use AI for, a rollout order that works, and the Indian policy context including CBSE's AI subject and NCERT's DIKSHA content.
The seven places AI actually helps a school
AI is most useful in schools for seven jobs: planning lessons and making question papers, giving first-pass feedback on written work, generating extra practice for students who are behind, handling attendance and fee and timetable admin, spotting students at risk of dropping out, supporting students with disabilities through speech and text tools, and answering the hundred repeat questions parents ask the front office. Notice what is not on that list. Replacing the teacher, and deciding a child's final marks.
That distinction is the whole game. AI is good at producing a rough draft fast and bad at being accountable. So you use it wherever a human will check the output anyway, and you keep it away from anything that goes on a report card unreviewed.
A teacher handling five sections of Class 8 Science spends real hours on things that are not teaching. Writing three versions of a unit test so neighbours cannot copy. Rewriting a chapter summary at two reading levels. Drafting the remedial worksheet for the eleven children who scored under 40 percent. Each of those is a twenty minute job that a chat tool turns into a four minute job plus a two minute check. Do that across a week and you have bought back an afternoon.
| Use case | What AI does | What the human must still do |
|---|---|---|
| Lesson planning | Drafts plans, worksheets, question papers at multiple difficulty levels | Check factual accuracy and syllabus alignment |
| Grading written work | First-pass feedback on structure, grammar, coverage of key points | Award the actual marks, handle borderline cases |
| Remedial practice | Generates extra questions targeted at a student's weak topics | Decide who needs remediation and sit with them |
| Admin and office | Attendance summaries, fee reminders, timetable clashes, circulars | Approve anything sent to parents |
| Early warning | Flags attendance and marks patterns linked to dropout risk | Home visit, counselling, the actual intervention |
| Accessibility | Speech-to-text, text-to-speech, live captions, translation | Verify accuracy for the specific student |
| Parent queries | Answers repeat questions on holidays, fees, timings, syllabus | Escalation path to a real person |
What this looks like on a Tuesday, not in a brochure
Take grading. A Class 9 English teacher gets 45 essays. She uploads a batch, gives the tool her rubric in plain words (content 5, organisation 3, language 4, spelling and neatness 3), and gets back a comment block per essay flagging thesis, paragraph structure and repeated grammar errors. She reads each one, overrides maybe a third, and spends her saved time on the eight children who genuinely cannot structure a paragraph. Marks stay hers. Research on automated essay scoring shows machine scores agree with human scores reasonably well on structure and mechanics and poorly on originality and argument, which is exactly why the teacher stays in the loop. If you want the mechanics of how these scoring systems work, see A Neural Approach to Automated Essay Scoring: What the Paper Says and How to Build One.
Take remediation. Most Indian classrooms have a spread of four or five grade levels sitting in one room. ASER 2023 found that only about 43 percent of 14 to 18 year olds could solve a simple three-digit by one-digit division problem. A teacher cannot write five worksheets a night. A tool can generate them in minutes, at the exact topic and level she names, in Hindi or Marathi or English. Adaptive practice systems take this further by adjusting the next question based on the last answer, which is the core idea behind Intelligent Tutoring Systems: How AI Actually Teaches One Student at a Time.
Take the office. Fee follow-up, transfer certificate drafts, bus route messages, exam datesheet circulars in two languages. This is where the fastest, least controversial wins sit, and it is where I would start in a school that has never used AI. Nobody's learning is at stake if a circular draft needs editing.
Where AI in schools goes wrong
Three failure modes repeat.
The first is hallucinated content reaching students. A tool will confidently produce a wrong date, a garbled formula, or a made-up quote from a poem. Any content that goes into a child's hands needs a teacher's eyes first, every time, no exceptions for a busy week.
The second is student data leaking. Typing a child's name, marks, health notes or home situation into a free public chatbot is a bad idea, and under India's DPDP Act 2023, processing children's data needs verifiable parental consent and prohibits tracking or targeted advertising to them. Practical rule: strip names before you paste. Use roll numbers or write "a Class 7 student" instead.
The third is students outsourcing thinking. Homework written by a chatbot teaches nothing. Detectors are unreliable and falsely accuse students, especially those writing in English as a second language, so do not build policy on them. Better design: more in-class writing, oral defence of submitted work, drafts shown at stages, and assignments that ask for something specific to the student's own town, family trade or last field trip. Make the work hard to fake instead of trying to catch fakes.
One more thing schools get wrong. Buying a platform before training a single teacher. Unused licences are the most common AI expense in Indian schools right now.
A rollout order that works in an Indian school
You do not need a budget line to start. You need one teacher, one free tool, and a term.
Start with teacher productivity, not student-facing tools. Pick three volunteer teachers across subjects. Give them a narrow task each for one month: question paper variants for one, worksheet differentiation for another, parent communication drafts for the third. Ask them to log time saved and mistakes caught. At the end of the month you have your own evidence, in your own school, from your own staff, which beats any vendor demo.
Then write a one-page policy. What data never gets typed into a tool. Which tools are approved. What students may and may not use AI for, by class level. Who checks AI-generated content before it reaches a child. One page, in the staff room, signed off by the principal.
After that, move to student-facing use. Practice generation and doubt-solving in supervised lab periods first, homework later. And run one admin automation in parallel, because that is the piece the management committee will actually notice.
- Month 1: three teachers, three narrow tasks, one free tool, written log of what worked
- Month 2: one-page AI policy covering data, approved tools, student use and mandatory teacher review
- Month 3: a 90-minute hands-on staff workshop run by your own three teachers, not an outside trainer
- Month 4: student-facing practice tools in supervised periods, plus one admin automation (fee reminders or attendance summaries)
- Ongoing: quarterly review of what is being used and what has quietly died
The Indian context: what is already in place
CBSE has offered Artificial Intelligence as a skill subject since 2019, from Class 8 onwards, and lakhs of students have taken it. NCERT content sits on DIKSHA with QR codes printed in textbooks. The NEP 2020 names AI in school education explicitly. So the policy cover exists. The gap is teacher confidence, not permission.
Schools in Kerala, Delhi and Andhra Pradesh have run AI-linked initiatives at the state level, from AI labs in government schools to teacher training modules. What separates the schools where it stuck from the ones where it did not is not funding. It is whether a couple of teachers found it genuinely useful for their own Sunday-night workload, and told the staff room.
So here is the honest answer to the question. AI in schools works when it takes drudgery off teachers so they can do the part machines cannot do, which is noticing that the quiet boy in the third row has stopped writing anything at all. Start there. Start this month, with one teacher and one task.
FAQs
1. Which AI tools can teachers use for free?
Free tiers of ChatGPT, Google Gemini and Microsoft Copilot handle most teacher tasks: lesson plans, worksheets, question paper variants, rubrics and parent messages. For Indian syllabus content, DIKSHA and NCERT's digital resources are free and government-backed, and Khan Academy's Khanmigo is available free to teachers in some regions.
2. Should schools ban ChatGPT for students?
Blanket bans mostly push usage underground and are hard to enforce on personal phones. A better approach is class-wise rules: no AI for first drafts in primary and middle school, supervised use for research and revision in senior classes, plus assessment designs like in-class writing and oral defence that make copied work obvious.
3. Can AI detect if a student used ChatGPT for homework?
Not reliably. AI detectors produce false positives, and OpenAI withdrew its own detector in 2023 for low accuracy, with non-native English writers flagged wrongly more often. Do not discipline a student on detector output alone; ask them to explain their work instead.
4. Is it legal to put student data into AI tools in India?
The Digital Personal Data Protection Act 2023 treats children's data as sensitive, requires verifiable parental consent for processing, and bars tracking and targeted advertising aimed at children. Practically, avoid entering student names, photos, health details or family information into public chatbots, and anonymise before you paste.
5. How much does it cost a school to start using AI?
Zero to start, if you use free chat tools for teacher work. Paid platforms for adaptive student practice or school ERP with AI features typically run into per-student annual fees, so validate value with free tools for a term before committing budget.
6. Does AI replace teachers?
No, and the evidence so far points the other way: AI shifts teacher time from preparation and correction toward instruction and pastoral care. It cannot manage a classroom, notice a distressed child, or be accountable for a wrong answer given to a student.