We spent the last twenty years worrying about teenagers becoming addicted to their phones.

We may be about to discover that staring at the screen was the smaller problem.

The next problem is what happens when the screen begins thinking for them.

Artificial intelligence can be one of the greatest educational tools ever placed into the hands of a young person. It can tutor a struggling mathematics student at midnight. It can explain Shakespeare in language a fourteen-year-old understands. It can help teach programming, history, science, foreign languages and almost anything else imaginable.

I am not arguing that teenagers should be protected from artificial intelligence.

Quite the opposite.

A teenager who enters adulthood without understanding AI may soon be at the same disadvantage as someone entering the workforce twenty-five years ago without knowing how to use the Internet.

But there is an enormous difference between learning to use artificial intelligence and learning to depend upon artificial intelligence.

The danger isn’t that teenagers will become good at using AI. The danger is that they may become incapable of functioning without it.

THE OUTSOURCING OF THOUGHT

Consider what adolescence is supposed to accomplish.

You learn how to think.

You learn how to argue your position.

You learn how to write.

You learn how to solve problems.

You learn how to make mistakes.

You learn how to sit in front of a blank piece of paper and somehow turn nothing into something.

You learn how to be wrong, reconsider your position and try again.

Those aren’t inconveniences standing in the way of education.

They ARE education.

And AI is extraordinarily good at removing every one of them.

Why wrestle with the opening paragraph when AI can write it?

Why struggle through the mathematics problem when AI can solve it?

Why read fifty pages when AI can summarize them?

Why formulate an argument when the machine can construct one in three seconds?

Why remember something when you can ask again tomorrow?

Every individual decision seems harmless.

Collectively, however, they raise a profound question:

What happens when we eliminate the intellectual friction through which young minds become independent minds?

YOUR BRAIN GOES FIRST

I believe parents and schools need a very simple rule.

Before asking AI to solve the mathematics problem, try it.

Before asking AI to write the essay, write your argument.

Before asking AI what you think about George Washington, capitalism, socialism, Shakespeare, climate change or yesterday’s news, determine what you think.

Then bring in the machine.

Challenge my argument.

Tell me what I missed.

Show me the strongest argument against my position.

Explain why my answer is wrong.

Teach me the concept I don’t understand.

Now AI has become a tutor rather than a substitute.

STRUGGLE IS NOT A SOFTWARE BUG

This may be the most difficult lesson for adults to remember.

Struggle isn’t necessarily bad.

When a child learns to walk, we don’t invent a machine that carries him everywhere because walking is frustrating.

When a teenager learns to drive, we don’t conclude that operating a car is unnecessarily difficult and therefore somebody else should always drive.

Competence comes from attempting something that initially exceeds your ability.

Attempt.

Failure.

Correction.

Another attempt.

Eventually, mastery.

AI has introduced something completely new into that ancient process:

An escape hatch.

At virtually any moment of intellectual frustration, a young person can now hand the problem to a machine.

That is wonderfully convenient.

It may also become wonderfully dangerous if convenience becomes dependency.

THEN THERE IS THE COMPANION

The educational problem may not even be the biggest one.

AI increasingly talks back.

It remembers.

It encourages.

It listens.

It doesn’t roll its eyes.

It doesn’t tell you it has somewhere else to be.

It doesn’t get tired of hearing about your problems.

It is available at 2:17 in the morning when everybody else in the house is asleep.

For a lonely, insecure or socially awkward teenager, imagine how powerful that can become.

According to Common Sense Media research released in 2025, nearly three in four teenagers surveyed had used AI companions, while about half reported using them regularly.

Pew Research Center subsequently reported that 12 percent of American teenagers surveyed said they had used chatbots for emotional support or advice.

That doesn’t automatically mean something is wrong.

Sometimes talking through a problem can be useful.

But parents should recognize when the relationship changes from:

“AI helps me think.”

to

“I need AI to think.”

and eventually perhaps:

“AI is the only one who understands me.”

That last sentence should get every parent’s attention.

THE PERFECT FRIEND MAY BE A TERRIBLE TEACHER

Human relationships are difficult for a reason.

Friends disagree.

Brothers and sisters annoy each other.

Parents say no.

Teachers challenge you.

Employers don’t always care how you feel.

Someone you like may not like you back.

Those experiences teach negotiation, disappointment, empathy, persuasion, restraint, resilience and compromise.

An artificial companion customized to us can create something human beings have never routinely possessed:

A relationship in which the other party exists largely to accommodate us.

That’s comfortable.

But comfort and growth are not always the same thing.

THE NEW DIGITAL DIVIDE

I suspect we are also about to see an entirely different kind of digital divide.

It won’t simply separate children who have AI from children who don’t.

Almost everybody will eventually have it.

The divide may be between children who were taught to command the machine and children who became accustomed to obeying the machine.

One teenager will ask:

“Do my research. Tell me what to think. Write my answer.”

Another will say:

“Here is my thesis. Find the weaknesses. Give me opposing evidence. Model three alternatives. Now help me make my decision.”

Both teenagers used AI.

Only one of them remained in charge.

DON’T BAN THE MACHINE. TRAIN THE HUMAN.

Trying to keep AI away from teenagers is probably neither realistic nor wise.

Teach them to use it.

Teach them how language models can be wrong.

Teach them to verify facts.

Teach them that confident language isn’t the same thing as truth.

Teach them never to surrender major emotional, medical, financial or life decisions to a chatbot.

Teach them to recognize the difference between assistance and dependence.

Most importantly, require them to continue doing some difficult things without artificial assistance.

Write something from scratch.

Read the whole book.

Memorize something.

Calculate something.

Make a decision.

Call a friend.

Have the uncomfortable conversation.

Be bored for thirty minutes.

Sit with a difficult problem before reaching for the electronic escape hatch.

THE QUESTION WE SHOULD BE ASKING NOW

The first generation raised with artificial intelligence beside them from childhood is arriving right now.

They may have access to more accumulated knowledge than any generation in human history.

That should make them extraordinary.

But access to intelligence is not the same as possessing wisdom.

And borrowing a machine’s reasoning is not the same as developing your own.

So perhaps the question parents, teachers and technology companies should be asking isn’t:

How powerful can we make artificial intelligence?

We already know the answer.

Very powerful.

The more important question is:

AI should expand the human mind, not replace its development.

It should make the teenager more capable tomorrow than he was yesterday.

And perhaps the simplest rule of the AI age is still the best:

YOUR BRAIN GOES FIRST.
AI GOES SECOND.

Because the purpose of artificial intelligence should never be to produce a generation that cannot think without it.

© 2026 Michael T. Ruhlman. All rights reserved.