When 13-year-old Riya was assigned a short story to read as part of an assignment, she photographed the page, fed it to an AI, read the summary in 40 seconds, wrote her paragraph, and went to bed. She never read the story. She would tell you, honestly, that she did not need to. She is right. But that is exactly the problem.

The companies building AI tools for education are not lying when they say their products open up learning. A child in a village with no books, no coaching centre, no college-educated parent can now have a difficult concept explained at midnight with more patience than any teacher exhausted by 40 students can manage. That matters enormously. But giving a child access to answers is not the same as building a mind that knows what to do with them. The AI industry has very little incentive to explain that difference.

Before AI came

Reading was already in trouble before AI arrived. For years, schools had quietly replaced sustained engagement with complex texts with excerpts, summaries and test-friendly passages that asked students to identify the main idea rather than live inside one. A study of English major students in the U.S. found that only five in 100 could correctly interpret a passage from Dickens. AI did not start this crisis. It arrived into one already in motion and handed an entire generation the perfect reason to stop trying.

What reading does to a young mind is not well understood even by people who love books. Children cannot comprehend what they read if they lack the background knowledge that texts quietly assume. That knowledge does not come from a summary or a chatbot. It comes from reading itself, from thousands of hours of genuine encounter with language, ideas, characters and arguments, built slowly, year by year. You cannot shortcut the infrastructure. When you try, you get students who can produce a paragraph about a story they never actually read. Everyone will call it education.

Teachers already know this. Thirty assignments arrive on their desks, fluent and identical in tone from students who cannot explain their own work when asked to do so. The thinking simply never happened. A 2025 Microsoft study reviewed by the Brookings Institute across 400 global research papers found that heavy AI use was measurably linked to weaker critical thinking and independent judgement. This is not simply a technology problem. Schools adopted the tools before they built the habits that would make those tools safe to use.

Way forward

The AI industry will not fix this on its own. Frictionless access to information is the product. That productive struggle is essential to learning does not appear in any roadmap or school adoption pitch. That conversation has to begin inside schools, inside homes, at kitchen tables, in classrooms where a teacher decides that reading is not optional, not today, not on her watch.

The schools getting this right are not doing anything complicated. Reading gets a protected period each day, treated with the same institutional seriousness as Maths. Assignments ask students to connect a text to their own experience, to disagree with it in their own voice, to produce something that cannot be photographed and fed into a machine. The text comes first. Every single time. Any tool, if it enters at all, enters only after the student has already thought.

A student who reads deeply and then uses AI arrives at a place that neither the book nor the machine could have taken her. She brings knowledge to the tool and questions the output because she has enough of her own thinking to notice when something is wrong. Reading gives her a spine. AI gives her reach. Without the first, the second is simply a very confident way of knowing nothing at all.

The library shelves are still full. The decision to build a child’s mind before handing her a shortcut is still available. But it will not wait for us indefinitely.

The writer is Assistant Professor, School of Holistic Development, MIT-ADT University, Pune.

Published - August 02, 2026 10:00 am IST