AI Isn't Making Kids Dumber. The Order Is.

Originally written in Traditional Chinese: AI 不是讓孩子變笨，是順序讓孩子變笨

In early September, New York City, the largest school district in the United States, announced a one-year pause on student-facing generative AI from kindergarten through eighth grade.

The same month, at an education forum in Taiwan, a National Taiwan University professor said that roughly 70% of secondary school students already use AI to do their homework.

One side is pausing everything.

The other is already at 70%.

Parents keep asking me: which side should we follow?

My answer: neither, not yet.

The question isn't whether kids use AI.

It's the order.

Let me start with a moment I still think about.

I had just finished a talk at a junior high school. A student came up and asked me, "Will AI replace me someday?"

His classmate said, "Just ask ChatGPT."

So he took out his phone. And asked.

A question about whether he would be replaced, handed to the very thing most likely to replace him.

It was a funny moment. But it stayed with me.

Not because he asked AI.

Because he skipped thinking about it himself, even for a second.

In 2025, a team led by Hamsa Bastani at the Wharton School published an experiment in PNAS.

Nearly 1,000 high school students in Turkey practiced math in three groups:

  • One group without AI.
  • One group with standard ChatGPT.
  • One group with an AI tutor designed by teachers not to hand out answers.

During practice, the standard ChatGPT group scored 48% higher than students without AI.

Looks great, right?

Then came the exam. AI removed.

That group scored 17% lower than the students who never used AI.

The AI tutor group? 127% higher during practice, and on the exam, about the same as students without AI. No drop.

Same kind of technology.

The difference wasn't the tool.

It was when AI came in, and what thinking it took over once it did.

I call this principle Human-First, Machine-Second.

When the goal is to learn, the human goes first.

When the goal is to get something done, the machine goes first.

Scheduling, cleaning up meeting notes, fixing formatting: that's getting things done. Hand it to AI, the faster the better.

Understanding a new idea, working out your own answer: that's learning. The brain has to show up first.

AI isn't making kids dumber.

The order is.

I want to be clear here, or this becomes just another feel-good post.

I understand New York's decision.

But most families in Taiwan can't enforce a total ban.

Bans push kids to use AI in secret. And secret use usually happens in the worst possible order.

What I want isn't less AI.

It's AI in a different order.

Here's something you can start tomorrow. I call it the 30-second snapshot.

Next time you, or your child, are about to open AI, first write four lines on paper:

  • The question: what do I want to ask?
  • What I already know: what do I already know about this?
  • My guess: what do I think the answer is?
  • Where I'm stuck: what exactly don't I understand?

Then paste those four lines into AI along with the question, and add one sentence:

"Please don't give me the answer. Guide me with questions, one question at a time."

In my course, I call this sentence the Socratic Coach.

What comes back is no longer a ready-made answer.

It's someone thinking alongside you.

I still remember that student turning to his phone.

I don't want kids to be banned from turning to AI.

I want them to turn to themselves first. Just once. Thirty seconds is enough.

When was the last time your child stopped to think before asking AI?

Get the Human-First practice card

Four steps, ten minutes, with copy-ready AI prompts. The card and my weekly letter are currently written in Traditional Chinese.

👉 Subscribe free to 碩學苑 (Traditional Chinese)

Follow in English

I also post in English on LinkedIn.

👉 Follow Haishuo Lee on LinkedIn

References