AI Will Do Everything. So What Do I Teach My Kid?

I write software for a living, and over the last four years I have watched AI quietly take over one part of my job after another — writing code, then compiling it, then testing and even deploying it. I’ve made my peace with that as a developer. What I have not made my peace with is the question it hands me as a parent.

Because if a machine can already do so much of what I spent years learning to do, the obvious thing to ask is: what on earth do I teach my kid? Why sit them down with math drills and spelling lists and, eventually, a programming course, if the future is a world where AI just… does all of it?

I don’t have a tidy answer. But I’ve thought about it more than almost anything else, and I’ve come to believe the question itself is a trap. Here is how I’ve come to think about it.

Note: This is a personal opinion piece from a developer and parent, not professional education or career advice.

The trap in the question

“AI will do everything, so what’s the point?” assumes the target is standing still. It isn’t. I am not preparing my child for a job — I’m preparing them for a life that will run sixty or seventy years, across a technology landscape that will reinvent itself many times over.

Think about the timescale. When I was a kid, “learn computers” meant something that is now almost unrecognizable. The specific skill I was taught was obsolete before I finished school. What wasn’t obsolete was the underlying habit: figuring out a new tool, breaking a problem into pieces, being unafraid of a blank screen. If I had been trained only for the tool of the moment, I’d have been retired at twelve.

So the honest reframe is not “what job survives AI?” It’s “what holds its value across a lifetime of tools I can’t yet imagine?” And on that question, the data is surprisingly reassuring.

What the numbers actually say

The World Economic Forum’s Future of Jobs Report 2025 — a survey of more than 1,000 employers representing 14 million workers — is about as close as we get to a consensus forecast. Its headline numbers sound scary at first: by 2030, 22% of today’s jobs will be disrupted, 92 million roles displaced. But 170 million new ones are created, for a net gain of 78 million jobs. Nearly two-fifths of the skills a job requires will change in five years.

That last figure is the one I keep coming back to. It doesn’t say “your skills will vanish.” It says they will churn — constantly. Preparing a child for that world means preparing them to keep changing, not to arrive pre-loaded with the right answers.

And here is the part that surprised me. When the WEF ranks the skills rising fastest toward 2030, the top of the list is exactly what you’d expect — AI and big data, cybersecurity, technological literacy. But immediately below them sit skills no machine is close to owning:

Creative thinking (#4)
Resilience, flexibility and agility (#5)
Curiosity and lifelong learning (#6)
Leadership and social influence (#7)
Analytical thinking (#9)

Meanwhile the skills in steepest decline are the rote ones: reading, writing and math as pure recall; manual precision; plain dependability and attention to detail. Read that list again. The things a machine now does better — memorization, mechanical accuracy — are falling. The things that make us human — curiosity, judgment, working with other people — are climbing. The data is quietly telling parents where to aim.

The three things I’m actually betting on

So when I strip away the anxiety, here is what I’ve decided to invest in with my own kid. Not a curriculum — three durable bets.

1. Learning how to learn. If the half-life of any specific skill is collapsing, then the meta-skill — the ability to pick up something new, fast, without being taught — is the only thing that compounds. This isn’t soft-headed. “Curiosity and lifelong learning” is literally #6 on the employers’ list. I care far more that my child enjoys the feeling of not-yet-understanding something than that they memorize any particular fact. A curious kid rebuilds themselves every few years for free. A kid trained only to follow instructions has to be re-trained by someone else, forever.

2. Judgment — knowing what’s good. This is the one my day job taught me. Modern AI produces an infinite supply of plausible output: plausible code, plausible essays, plausible plans. What it cannot do is decide which of them is actually good, actually true, actually worth shipping. In my own work, as the machine climbed each rung, the scarce thing that stayed mine was always the same: the moment of looking at the output and saying “yes, this — not that.” Taste. Discernment. The ability to be a good editor of a machine that will out-produce you a thousand to one. I want my kid to be that editor, not that producer.

3. Being human, with other humans. Empathy, communication, the ability to read a room and move a group of people — these are the skills the futurist Sinead Bovell nicely calls the ones “we can foster for free.” They cost nothing and they are the least automatable things on the list. In a world where content is infinite and cheap, trust between real people becomes the scarce commodity. I’d rather my child be the person others want in the room than the person who can recite the most facts in it.

But should I still teach them to code?

I’m a developer, so this is the question I actually get asked at dinner parties. My honest answer: yes — but not for the reason you’d think.

Not so my kid can become a fast typist of syntax; the machine already wins that. I’d teach coding the way I’d teach chess or a musical instrument — as a gym for a particular kind of thinking. Coding forces you to break a fuzzy problem into exact steps, to confront a bug that doesn’t care about your feelings, to learn cause and effect with brutal honesty. That cognitive workout survives even if no human ever hand-writes a for loop again. As the field itself is discovering, the future of programming is shifting from typing code toward architecture, judgment, and knowing what to build. Which is really just bet #2 again, wearing a hoodie.

What I’m deliberately not doing

I’m not chasing the hot skill. Two years ago “prompt engineering” was going to be the career of the century; today it’s a footnote, half-absorbed by better models. If I optimize my kid for today’s AI, I’m training them for a tool that will be obsolete before they graduate.

This is why I’ve stopped asking my child the classic question, “what do you want to be when you grow up?” Bovell puts it better than I can: we should decouple jobs from identity and instead ask kids what problems they want to solve. A job title is a snapshot of a labor market that will have moved on. A problem worth solving is a compass that keeps pointing somewhere real no matter how the tools change.

The bottom line

Here’s the thing I finally understood. “AI will do everything” is the wrong fear, because everything is not the same as anything that matters. A machine can do tasks. It cannot want an outcome, care whether it’s right, feel responsible for who it affects, or decide what is worth doing in the first place. That entire domain — wanting, caring, judging, choosing — is untouched, and it’s exactly where I’m raising my kid to live.

I’m not trying to build a child who can out-compute the machine. That race is already lost, and it was always the wrong race. I’m trying to build the person who knows what’s worth computing — and who can look at an infinite stream of machine-made answers and still say, with a human’s judgment and a human’s conscience, “yes, this one.”

That skill has no expiry date. As a developer watching the line move up my own ladder year after year, it’s the one rung I’ve never seen a machine take. It’s the one I’m handing to my kid.

FAQ

If AI can do most jobs, is it pointless to teach kids specific skills? No — but the purpose of teaching them changes. Specific skills (like coding or math) are worth learning less as job training and more as exercise for durable capacities: problem-solving, logical thinking, and judgment. The WEF’s data shows the fastest-declining skills are rote ones (memorization, mechanical accuracy) while human skills like creativity, curiosity and analytical thinking are rising.

What are the most future-proof skills for children? Across the WEF Future of Jobs 2025, OECD education research, and most futurists, the recurring answers are: curiosity and the ability to learn continuously, creative and critical thinking, judgment/discernment, resilience and adaptability, and human skills like communication and empathy. These stay valuable precisely because they are the hardest to automate.

Should my child still learn to code in the AI era? Many developers, myself included, say yes — but as cognitive training rather than vocational training. Coding teaches problem decomposition and rigorous cause-and-effect thinking, which remain valuable even as AI writes more of the actual syntax. The work is shifting from writing code toward deciding what to build and judging whether the result is good.

Why stop asking kids “what do you want to be when you grow up?” Because a job title assumes a stable labor market that no longer exists. Futurist Sinead Bovell suggests asking what problems a child wants to solve instead — a framing that survives changes in technology and doesn’t tie a child’s identity to a specific role that AI might reshape or absorb.

Sources

This article may contain affiliate links. Nothing here is financial or investment advice.