Should Colleges Ban AI? What Actually Helps Students Learn

Arthi's Take · By Arthi Rajendran, The Arthi AI Collective · Published · Updated

This morning I read Anurag Behar's column in Mint, and I could not stop thinking about one paragraph.

He describes a loop. A teacher asks a machine to prepare the lesson. A student asks a machine to write the answer. Another machine marks it. In his words: "Nowhere in this chain has any human mind learnt anything."

He is right. And yet I disagree with where he lands. Let me explain why, respectfully, because this argument matters to every student and parent I work with.

What Anurag Behar wrote

Anurag Behar is the CEO of the Azim Premji Foundation, and he has spent years thinking about Indian education. His column, "Why artificial intelligence could be the last blow to higher education in India", was published by Mint on 11 November 2026. It is a Mint Premium article.

Most of the piece is not about AI at all. He argues that India's higher education is in worse shape than its school education, and gives five reasons:

  • It is harder to decide what colleges should teach than what schools should.
  • Institutions have tied teaching to research, and very few do both well.
  • There is a tacit compact between teachers and students not to take classes seriously.
  • Colleges resist change, and are good at it.
  • Public funding has not kept pace with enrolment, and commercialisation has filled the gap.

Then he calls AI "a bigger threat than all of this." Teachers use it to teach, students to respond, and teachers again to assess. He criticises "reckless and unthinking leaders" who call for more and more AI. He is careful to say there are good institutions and great teachers, estimating that 10 to 15 percent of institutions are serious about their work. And he ends by calling for "(mostly) banishing AI from education," along with hard work to rebuild the foundations.

His closing line is the one I agree with most: if we fail, "the cost as ever will be borne by those who have little today and need good education the most."

Where I agree with him

The empty loop is real. When a lesson plan, an assignment and its grading are all produced by machines, the college has become a very expensive way to move text from one AI to another. Nobody learnt anything, and everyone got a mark.

I also agree that AI did not create this problem. Behar's five reasons come first in his piece for a reason. If students and teachers had already stopped showing up, AI simply made the emptiness faster and harder to see.

And I agree that "more AI" is not a strategy. Buying tools for a college that has not decided what students should learn will not fix anything.

Where I disagree: banning AI is the wrong fix

Here is the problem with a ban. The loop he describes has a weak link, and it is not the AI. It is the assessment.

If a machine can write the answer and another machine can mark it, the assignment was never measuring understanding. It was measuring the ability to produce a certain kind of text. AI just exposed that.

A ban also does not reach where the use happens. Students use AI on their own phones, at home, late at night. A ban in the classroom does not stop the loop. It only stops the honest conversation about it.

And it sends graduates into a world that is moving the other way. On the same day as Behar's column, The Economic Times reported that IIM Kozhikode, IIM Lucknow, ISB and BITSoM are reworking their MBA programmes for AI. At BITSoM, every MBA student is required to build an AI agent. IIM Kozhikode, the report says, is rethinking what assignments should test. And Jun Yang, dean of Nanyang Business School in Singapore, is quoted in the same report: "AI is not making the MBA obsolete. It is making an MBA built primarily around knowledge transmission obsolete."

That last line is the whole argument. The answer is not less AI. It is less of the kind of education that AI can fake.

Finally, I worry about who a ban hurts. Students from families and schools with resources will learn to use AI well anyway, through private courses, internships and parents who use it at work. A student at an under-resourced college, the very student Behar is worried about, is the one most likely to graduate without that skill if their college pretends AI does not exist.

What actually helps students learn

1. Change how learning is assessed

Assessment is the lever. A few things work, and many Indian colleges already know them:

  • Bring back the viva. A five-minute conversation about a submitted assignment shows quickly whether a student understands it. AI cannot sit the viva for them.
  • Do more writing and problem solving in class. Short, regular, handwritten or offline tasks, not just one big take-home submission.
  • Grade the process, not only the product. Ask for drafts, notes and, where AI was used, the prompts and what the student changed and why.
  • Use local and personal material. An analysis of your own town's water supply, a family business, or data the class collected is harder to outsource than a textbook essay.
  • Ask students to critique AI. Give them an AI answer with mistakes and ask them to find and fix them. That needs real understanding.

2. Teach AI literacy openly

AI literacy is not "how to write a prompt." It is a set of habits:

  • Knowing when AI helps and when it replaces the thinking you are supposed to be doing
  • Checking every factual claim against a real source
  • Noticing confident mistakes and invented references
  • Keeping private and personal data out of AI tools
  • Disclosing honestly when and how AI was used
  • Being able to explain and defend any submitted work in your own words

I wrote more on the thinking side in how to use ChatGPT without switching off your brain and how to stop AI making your decisions.

3. Support teachers, not just students

Teachers are part of the loop too. Many are overloaded, and AI lesson plans are tempting. Instead of pretending that does not happen, colleges can set a simple rule: use AI to save time on preparation, but the teacher reviews every line, and the teacher marks student work. My guide on AI for teachers and colleges has practical ideas.

4. Set clear rules for each course

"No AI ever" and "use anything" are both easy to write and hard to live with. A better policy says, for each assignment, what is allowed: AI for brainstorming but not drafting, AI for code but with an explanation, or no AI for this one because the point is to practise the skill yourself.

If you are a student or parent right now

You do not have to wait for your college to fix this.

  • Students: use AI as a tutor, not a ghostwriter. Ask it to quiz you, explain a step you missed, or check your reasoning after you have tried. Before you submit anything, ask yourself: could I explain this to my teacher without looking at it?
  • Parents: ask your child's college what its AI policy is and how it checks understanding. Ask your child to walk you through one assignment out loud. If your college-going child needs more support, my AI Launchpad for Students program is 1:1 mentorship for college students to build portfolio-ready AI projects.
  • Early career professionals: if your degree did not teach you this, learn it now. Employers will expect you to use AI and to know when its answer is wrong.

The goal is a human who learnt something

Behar and I want the same thing: a student who walks out of college having actually learnt to think. I just do not believe we get there by pretending AI is not in the room. We get there by making sure that whatever happens in the room, a human mind does the learning.

That is how I teach at The Arthi AI Collective. I help students and working professionals use AI as a thinking partner, while keeping the judgement their own. If you are a college student, or a parent of one, AI Launchpad for Students is 1:1 AI mentorship to build portfolio-ready AI projects and stand out in placements, built with a mentor, not from recordings. For every track, explore the programs.

Sources

Arthi Rajendran, The Arthi AI Collective

Frequently asked questions

Should colleges in India ban AI?

I do not think a blanket ban works. Students can use AI on their own phones outside class, and they will graduate into workplaces where AI is part of the job. A better fix is to change assessment so it shows what a student actually understands, and to teach AI literacy openly.

What did Anurag Behar say about AI in higher education?

In a Mint opinion piece on 11 November 2026, the Azim Premji Foundation CEO described a loop where a machine prepares the lesson, a machine writes the answer and a machine marks it, and wrote that nowhere in this chain has any human mind learnt anything. He ended by calling for (mostly) banishing AI from education.

How can teachers stop students using AI to cheat on assignments?

Design assignments where a pasted AI answer is not enough: short vivas, in-class writing, drafts and notes submitted along with the final work, and tasks built on local data or the student's own experience. Ask students to critique an AI answer rather than just produce one.

What is AI literacy for college students?

Knowing when to use AI and when not to, checking its answers against sources, spotting confident mistakes, protecting private data, disclosing AI use honestly, and being able to explain and defend any work you submit in your own words.

Are Indian business schools teaching AI?

Yes. The Economic Times reported on 11 November 2026 that IIM Kozhikode, IIM Lucknow, ISB and BITSoM are changing their MBA programmes for AI. At BITSoM, every MBA student is required to build an AI agent, and IIM Kozhikode is rethinking what assignments should test.

Work with Arthi

Explore 1:1 AI mentorship

Related posts