Side by Side: The Tutoring Room Drawn by AI and the One I Remember

Last night I typed a prompt into DALL-E. "Small tuition centre in a Delhi residential colony, early 2000s, ceiling fan, steel chairs, whiteboard." The moment the result appeared, I stared at the screen for a long time. It seemed right, and at the same time, not right at all. That strange feeling is why I'm writing this.

The tuition centre I attended was on the ground floor of a two-storey house, tucked inside the lanes of Laxmi Nagar. The teacher — we just called him "Sir" — had cleared out his living room to use as a classroom. The light green paint on the walls was half peeling, and the calendar hanging by the door was always two or three months out of date. One of the three ceiling fans was perpetually broken, and in summer the other two spun at full speed, scattering papers everywhere. I can still feel the sensation of pressing test sheets down with my knees to keep them in place while solving problems.

The Tuition Centre AI Showed Me

The image DALL-E generated was clean. Startlingly so. The steel chairs were right. The whiteboard was there. A fan hung from the ceiling. But everything was too orderly. The chairs were evenly spaced, there wasn't a single stain on the floor, and light streamed softly through the window. It looked less like a tuition centre and more like a café themed after one.

On my second attempt, I made the prompt more specific. "Worn-out steel chairs, peeling green paint, cluttered desk with old registers, dim tubelight." The result improved slightly. The walls did look worn. But the AI treated "worn" as an aesthetic — intentional vintage. Real wear isn't like that. The stack of old attendance registers and spent ballpoint pens on Sir's desk, the plastic bag the milkman had left in the corner, the dust wedged into the edges of the walls. These are accumulations of things nobody ever intended.

The Tuition Centre in My Memory

When I reach back into my memory, what comes first isn't sight but sound. The whistle of a pressure cooker from the lane, a news anchor's voice leaking from a neighbour's TV, the rhythm of Sir tapping his marker against the whiteboard — tak-tak — to stress key points. Then smell. Marker ink mingled with the chai Sir's wife brewed in the kitchen, so that even now, whenever I smell a marker, the taste of chai spreads across my tongue.

The chairs were folding steel chairs with low backs — lean back and it felt like you'd topple over. I actually did fall once. Maths class. I was solving a problem, zoned out staring at the ceiling, and lost my balance. I expected Sir to get angry, but he just said, "Fell again?" The word "again" told me I wasn't the only one.

What Memory Erases, What AI Fills In

Placed side by side, AI captures the structure of a space fairly accurately — the size of the room, the arrangement of furniture, the direction of light. But AI has no trace of time. The dent worn into a chair seat from thousands of sittings, the marker stain in one corner of the whiteboard that would never wipe clean, the tape residue on the wall from a multiplication table or timetable that once hung there — these things are hard to describe in a prompt. Even if you do, the AI places them as decoration; it cannot reproduce the random wear that time creates.

Then again, my memory isn't reliable either. I'm certain the walls were light green, but they might have been sky blue. Memory keeps favourite details sharp and lets the rest blur. I don't even know exactly how many of us sat in that room. Twelve? Fifteen? I remember the tone of Sir's voice, but the exact contours of his face are hazy. Memory is footage edited around sensation; an AI image is a stage set assembled around visuals. Neither is that tuition centre.

One more thing. In AI images, there are either no people or, at best, anonymous silhouettes. In my memory, that tuition centre has meaning entirely because of people. Priya, who sat next to me and would slyly show me her answers. Rohit, who always arrived five minutes late and got scolded by Sir yet never stopped grinning. The evenings before exams when we stayed until 9 p.m. copying each other's notes. We'd buy 30-rupee samosas from the lane-side stall and share them. There is no way to explain that taste to an AI.

Sometimes, walking through Delhi's lanes, I spot signs for similar tuition centres. There must still be hundreds of them. If the smell of markers drifts down from some second-floor room in some lane, I pause for a moment. I'll forget an AI-generated image by the next day, but that smell takes me back to age thirteen every single time. I know memory distorts. Yet the distorted memory still feels closer to the real thing — probably because it carries the smell, the sound, and the grease of a samosa within it.