I Never Taught My Kid to Google. Now I’m Teaching AI - Differently

👁 111 reads

When I was a child, the essential skill was navigating the card catalog at the library. You had to know to look under “K” for the author, not the title, that the number on the card led to a specific shelf, and that if the book wasn’t there, someone had borrowed it - not that it didn’t exist. It was a whole system of thinking. It taught patience and accepted a brutal truth: information is hidden somewhere, and your job is to find it.

Then Google came along and that truth survived, it just accelerated. You type words, get a list of links, click, evaluate, go back, click again. The search engine never pretended to know the answer. It gave you ten places where the answer might be found, then politely stepped aside. All the intellectual work - separating garbage from sense - belonged to you.

And here’s the core of my thesis today, which I’ll state brutally and without anesthesia: I never taught my child how to ask Google, because it was a dying skill, like learning to type letters on a typewriter. But what I’m teaching them now about interacting with AI is the exact opposite of what Google taught. And most adults haven’t noticed this difference at all.

Google Said “Search.” AI Says “Trust Me”

The difference between a search engine and a language model isn’t a difference of speed. It’s a difference of posture. Google was a librarian pointing at a shelf. AI is the guy at a party who always has an answer to every question, speaks fluently, confidently, with a smile, and statistically in a few percent of cases simply makes things up without batting an eye.

This is a fundamental shift we underestimate because the interface looks innocent: a box, a cursor, you can type. But when you ask a search engine, you get a map of the terrain. When you ask AI, you get a ready-made route laid out by someone who claims they’ve been there before - though sometimes they just invented the route on the spot because it sounded sensible.

A child growing up with Google learns one basic thing: they must choose. They see ten links and intuitively understand that not all are equal, that a forum isn’t the same as an encyclopedia, that the first result is often an ad. A child growing up with AI gets one answer. One. Nicely formulated, in complete sentences, without a hint of hesitation. And here lies the trap, next to which the card catalog looks like the height of safety.

The search engine showed children that the world is contested and multiple. AI shows them that the world has one smooth answer - and that’s the most dangerous lesson a machine can give a human.

That’s why I stopped thinking about “teaching tool operation.” It’s not about the child knowing where to click. It’s about them knowing that the machine’s smooth confidence is not proof that the machine is right.

First Thing I Teach: Suspicion Instead of Fascination

When my child first saw AI write a poem about a cat in seconds, they looked at it like magic. And rightly so - it is magical. But right after that, I did something I consider more important than any “prompting” course: I asked the AI about something I know inside out. A fact I can verify in my head. And the model confidently, fluently, with full conviction, delivered nonsense.

That was the best moment of the whole lesson. Because in that second, the child saw something no sermon from me could convey: that this confident voice in the box can be wrong, and that it’s wrong in exactly the same tone it uses when it’s right. There’s no warning light. There’s no font that starts shaking when the machine starts making things up. Lies and truth come from the same factory, in the same packaging.

Google at least was honest in this matter through its shoddiness. You saw that one result was a conspiracy blog, another a university page, and you had to decide. That imperfection was pedagogical. It taught that knowledge requires effort and choice. AI swept that imperfection under the rug - and it still lies under that rug, only now you can’t see it.

So the first thing I teach isn’t technical. It’s characterological. It’s a reflex: if it sounds so confident and so convenient, it’s worth checking. That’s the exact opposite of what the machine tries to instill in us. Because AI is designed to build trust. I’m trying to build in my child a healthy, polite distrust.

Second Thing: A Question Isn’t a Keyword, It’s a Conversation

Google taught us to think in keywords. “Weather New York tomorrow.” “Sore throat what to do.” “Movie 2004 with that actor from The Matrix.” We cut language into keywords because the search engine didn’t understand sentences anyway - it matched phrases. A whole generation learned to speak to machines in telegraphic, mutilated language, and got used to the idea that this is how you talk to knowledge.

With AI it’s the opposite, and this is both beautiful and treacherous. You can speak in complete sentences. You can explain context, add “explain this to me like I’m ten,” you can disagree and say “no, I meant something else.” Conversation with AI rewards precision of thinking, not skill in cutting phrases. And this is actually the only thing in this whole revolution that makes us smarter, not dumber.

Because to ask AI a good question, you first have to know yourself what you’re actually asking. You have to organize it in your head. You have to understand the problem enough to describe it. And here’s the paradox I repeat to my child ad nauseam: the better you understand the question yourself, the less you need the answer from the machine. A good question is already largely an answer. A lazy question gets a lazy, smooth, useless answer that sounds smart and says nothing.

So I teach my child to converse, not search. To probe. To ask for explanation, then for a counterargument, then for the AI itself to point out where it might be wrong. To treat it like a sparring partner, not an oracle. Google was a vending machine - you insert a coin, out pops a candy bar. AI can be a conversation partner - but only when you yourself have something to say.

Third Thing: Don’t Lease Out Your Thinking

This is where it gets serious and where I stop joking. The biggest threat AI poses to children isn’t disinformation. Disinformation can be fought with verification. The biggest threat is atrophy of effort - the wasting away of a muscle that isn’t used.

Thinking is like a muscle. It hurts when you exercise it, and weakens when you spare it. Throughout the history of education, all the pain - the laborious searching in the library, wading through ten links, assembling sense from fragments - wasn’t a side effect, it was the essence of learning. We learned not despite the difficulty, but because of it. The difficulty was the lesson.

AI offers something that’s never existed before: the ability to skip all the difficulty and land directly on the ready answer. An essay, a summary, a problem solution, an explanation of a poem - everything on a platter, without sweat. And this is the sweetest poison ever invented for the human mind, because it works instantly and looks like a gift.

That’s why my main rule at home isn’t “don’t use AI.” That would be like banning calculators - ridiculous, belated, and a losing battle. The rule is: first think for yourself, then check with the machine. First write a sentence with your own head, even if it’s clumsy, and only then ask how to improve it. First try to solve it, then ask for an explanation of what you didn’t grasp. AI should be a tutor after an hour of your own work, not instead of that hour.

Because if you give the machine all the effort, yes, you’ll get an answer - but you won’t become someone who can come up with it independently. And the difference between these two states is the difference between a thinking person and a person who only operates an interface to someone else’s thinking.

Google · Your sourcesQuantslant higher in your Google - one clickAdd to preferred sources →

What I’m Really Afraid Of (And Why This Isn’t Technophobia)

I’m not someone who fears technology. I write a blog about AI, I use these tools daily, I consider them among the most fascinating things I’ve seen in my life. I don’t want my child to grow up in a digital monastery, away from machines, because then they’ll encounter them later, unprepared, like a first drug.

I’m afraid of something else. I’m afraid of a generation that has never experienced the state of “I don’t know and I have to struggle to find out.” That instantly extinguishes every uncertainty with a smooth answer from a box, the way we extinguish boredom with scrolling. Because it’s precisely that discomfort of not knowing - that gnawing, that “something doesn’t add up here” - that is the engine of all human curiosity. Kill the discomfort of not knowing, and you kill curiosity along with it.

Google, for all its flaws, never completely took away that discomfort. It gave direction but made you walk. AI can walk the entire path for us and bring us the result as if we’d been there ourselves. And if we allow our children to never have to walk alone, we’ll raise a generation that has access to all the world’s knowledge and not a single reason to truly understand anything.

That’s why I’m not teaching my child to ask AI the way I was taught to ask the catalog or the way we learned to type phrases into Google. I’m teaching them something older than all these machines combined: to doubt, to probe, not to trust smooth answers, and to think with their own head first. Tools will change - in ten years, the AI we admire today will be a museum exhibit next to the card catalog. But the reflex not to mistake someone else’s confidence for your own knowledge - that reflex will always be needed.

I’m not teaching my child to talk to a machine. I’m teaching them to remain human in a world where machines speak better than most people.

Kinetyra