The Tutor Could Not Make Them Care

“Can you write an essay about the history of pizza?”

That was one of the questions the children asked the expensive new tutor.

Not help me understand fractions. Not why does this equation work. Pizza. A twelve-year-old, faced with a machine built by adults with whiteboards and funding rounds and a tragic confidence in their own nouns, did the only reasonable thing. He pressed the big glowing button and asked it for something useless.

I have more faith in that child than I have in the adults.

Eighteen middle schools in Tennessee gave students access to Khanmigo for two years. The arrangement was supposed to be the clean modern dream: a patient tutor in every pocket, available all day, unable to get tired, unable to snap at you for tapping your pencil, supposedly wise enough to guide you toward the answer without giving it away.

The machine stood there with its sleeves rolled up.

The kids mostly walked around it.

When they did use it, many tried to get it to cough up answers. Or they poked it with questions about pizza and jokes and whatever other small acts of vandalism a child can commit with a keyboard. Access was nearly universal. Engagement was thin. That sentence from the researchers has the bleak little rhythm of a janitor describing a school dance after everybody has gone home.

There will be men who read this and decide the children are the problem.

They always do.

They will say kids lack discipline. They will say the interface needs more sparkle. They will invent a badge system, a streak, a little animated owl with a badge and a whip. They will hold a meeting about improving adoption. Then somebody will suggest giving the machine a warmer voice, because nothing says education like a customer-service hostage recording.

Nobody will say the dangerous thing: a child is not a browser with poor settings.

I remember being twelve. The adults had already spent years explaining that education was the key to a door nobody could quite point to. Learn this now, they said, and later your life will be better. Later was a country without maps. The classroom clock moved by legal decree. A page of algebra could sit in front of you like a brick wall, and the person standing beside it could be kind, cruel, tired, brilliant, bored, or quietly falling apart. You noticed all of it.

That is what the tutors of my childhood had that no software brochure can fit in a bullet point: a body in the room.

A real teacher can see that you have not slept. They can hear the class go dead. They can make an offhand remark that turns a light on. They can fail you badly enough that you finally get angry. They can tell when the answer is copied, when the joke is covering fear, when the kid in the back who has never raised a hand knows exactly what is going on.

They can also be lousy. Let us not write their halos too fast. I had teachers who could drain the color out of a Tuesday. But even the lousy ones were there. They had breath. They had a bad mood. You knew somebody had to meet your eyes after you handed in the garbage.

A chatbot has manners because it cannot afford a personality.

That is its great limitation and its great sales pitch. It will wait forever. It will never look disappointed. It will produce another explanation in a different style, then another, then another, while a child learns the newer and cheaper lesson: if the work is unpleasant, keep asking the box until the unpleasantness disappears.

Maybe this is what the companies mean by personalized learning. Each child alone with a polite liar, rearranging the difficulty until it no longer resembles learning.

The Tennessee students who used the system did move faster in math than a control group. Fine. A machine can sometimes be useful. So can a calculator, a library, a woman down the block who knows how to explain a tax form without making you feel stupid. The honest question is not whether a tool can help. The honest question is what everybody else is about to stop paying for because the tool exists.

The study’s researchers landed on the answer the hard way: human attention is the key ingredient. They might as well have announced that water is wet and landlords enjoy money. But apparently this has to be discovered again whenever a machine arrives in a school wearing the fake mustache of progress.

The strongest results came when the technology helped human tutors, not when it was parked in front of students like a babysitter that had read every textbook and never once wanted anything.

The bill comes due in people.

The machine cannot make a kid care. It cannot make a kid believe the future has room for them. It cannot make a classroom feel like a place where failure does not become a permanent label stuck to your shirt. It can answer questions. It can generate a serviceable history of pizza. It can even keep its voice calm while being asked to say something filthy at 10:14 on a Wednesday.

But attention is not an output. It is a cost.

Somebody has to pay it in time. Somebody has to sit there long enough to notice a child has turned away from the work, and long enough to ask why without making them wish they had stayed quiet.

The kid who asked about pizza understood the situation perfectly. He had been handed a thing that wanted him to learn, and he asked it to entertain him instead.

The machine did not get angry. That was the problem.

Source: Scientists Gave 12-Year-Olds Access to an AI Tutor and You Probably Already Know What They Did With It