Hey folks, welcome back to Paradox Weekly, a Masters' Union newsletter, where we break down the ideas, trends, and contradictions shaping business today.

Before we dive in, a quick poll:

Three months ago, the Ministry of Education announced that AI and Computational Thinking will become mandatory subjects from Class 3 onwards, starting academic year 2026-27.

Since the announcement, I’ve been hearing all praises such as:
"Preparing kids for the future."
"Building an AI-ready workforce."

Meanwhile, I had a conversation with a professor last week who teaches economics at a tier-1 college.

He told me about a student who’d submitted an essay with perfect English. It was evident that the thought wasn’t his.

He knew it was AI.
The student knew he knew.

The professor asked him to explain one of his own arguments.
The student couldn’t.

He hadn’t actually engaged with the ideas.
Instead he’d just typed a prompt and cleaned up the output.

The professor had no recourse.

Detection tools flagged it at 40%.
Turnitin said “inconclusive.” 

The student will graduate.
And soon get a job.

This is happening everywhere.
And we’re about to teach this skill to 260 million more students.

Before we go further. Here’s what happened this week:

Fantastic to see so much effort by RISE & SBM students today. One of the students got a B2B order worth 1.5L. Another one launched a new peanut butter brand in ice cream flavours.

Hosted Anupam Mittal, the OG shark on campus. Stay tuned for what happened!

Back to the topic. 

Whenever AI in education comes up, I see two reactions:

The optimists say this is visionary. 

India is preparing students for the future.
AI literacy will be as fundamental as reading and math. 

The IMF estimates 40% of jobs globally will be affected by AI.
In richer economies, closer to 60%..

The skeptics see disaster.

97% of Gen Z students have used AI for schoolwork.

Professors assign essays, students feed prompts to AI, submit without reading.
When asked to defend their arguments orally, they can't.

Both miss the point.

AI is to make you think harder. But it’s making people stop thinking entirely…

MIT’s Media Lab ran a study with 54 students over four months.

Then they tracked brain activity with EEG while these students wrote essays.

The results were what I expected.

Students using ChatGPT had the lowest neural engagement.

The students writing unaided showed the strongest connectivity, most engagement, and had the best recall.

83% of ChatGPT users couldn’t recall key points from essays they had written minutes earlier.

They just wrote the essays and didn’t remember what they wrote.

Lead researcher Nataliya Kosmyna called it “cognitive debt.”
The more you outsource thinking, the less you encode into memory.

Two English teachers who assessed the essays called them “soulless”, as they were structurally perfect but intellectually empty.

While the study is a preprint with a small sample size, and is not peer-reviewed yet.
The pattern matches what every teacher I talk to describes.

Wharton ran a study with nearly 1,000 high school students in Turkey.

They also had a third group with “GPT Tutor.” 

An AI designed to only ask questions and never give answers.
This would force students to reason through problems themselves.

These students showed no harm compared to the controls.

Whether AI does the thinking or makes you think changes everything.

The simple answer is AI induced training.

A Harvard randomized trial found AI tutoring produced learning gains double the comparison group.

The reason is that AI never gave direct answers.
It asked questions, scaffolded their understanding and made students do the work themselves.

ChatGPT is optimized for speed.

Give me the answer.
Write the essay.
Skip the struggle.

Educational AI tools like NotebookLLM are optimized for learning. 

Guide me through.
Make me explain.
Don’t let me shortcut.

This is how I use it for instance:

First, it gives me a mindmap.

A structure that shows me what exists to be thought about.

I now have to click into branches, follow threads, and make sense of the relationships by going down the rabbit hole in an organised way (instead of me making this by reading and spending hours).

Then, synthesis.

Once I’ve gone through the material, I ask for a recap with an infographic.
This is so I can check if what I think I understood actually holds together.

And finally, recall.

I generate flashcards and test myself.

This is the hard part where you realise what you actually remember and what you were just nodding along to.

The issue is that India’s curriculum framework says AI should help with “lesson planning and resource design.” 

But nothing about teaching students when not to use it or rather how to use it.

It’s a large-scale mission.

260 million students across 1.5 million schools, and 10 million teachers who need training.

Classes start in seven months.

Roughly 50% of these schools still lack basic digital infrastructure.
And the government is exploring “unplugged learning” activities to teach AI concepts without computers.

We’re teaching AI literacy to students who might not have access to AI.

The employability pipeline is already broken.
Aspiring Minds Employability Report found that 80% of Indian engineers aren’t employable in the knowledge economy. 

Only 2.5% possess AI skills.
That was 2019. Before ChatGPT existed.

If AI becomes a crutch before students learn to walk, that employability gap gets worse.

India is running two experiments at once.

Both experiments are running with the same tool, curriculum, and a seven-month timeline.

The debate frames this as "AI versus education," but that's the wrong frame.

The real question is about struggle. 

Learning happens when you struggle with something, when you stare at a problem and don't know the answer, when you write a sentence, delete it, and rewrite it. 

That struggle is where the learning lives.

AI removes this.
It is simultaneously a feature and a bug.

The Ministry announcement says nothing about how to prevent the first scenario while enabling the second. 

It talks about curriculum and teacher training and digital content, but nothing about the fact that we're giving students a shortcut before they've learned what they're shortcutting.

I use AI every day. 

I also know that the struggle of figuring something out is where the learning happens.
The prompt that gets you a perfect essay is also the prompt that ensures you learn nothing.

We're about to teach that prompt to eight-year-olds. Or in other words: we're building an education system that promises to eliminate struggle. That should worry us.

At the end of the day, it’s going to depend on how motivated a learner is to learn. They will have all the shortcuts in the world, and they are the only ones who can choose to leverage them, or use these tools to understand in ways traditional schooling couldn't.

Hit reply: If you’re a teacher, student, or parent, I want to know what you’re seeing. Are students learning more? Thinking less? Both?

I read every email.

Until next week,
Pratham