I’ve been a software engineer for over eight years. I write code every day, I build AI systems for a living, and I’m raising kids who will enter a job market that looks nothing like the one I entered. At least once a week, someone — a friend, a client, a fellow parent at school — asks me the same question: “Should my kids learn to code?”
My answer has changed. Here’s why, and what I think actually matters.
The Question Everyone’s Asking
The anxiety is understandable. AI can now write code. Not just simple scripts — it can build full applications, debug complex systems, and explain its reasoning step by step. If AI can code, why would you teach your kids to do something a machine does better?
This question is based on a misunderstanding of what “coding” actually is. And it’s based on a misunderstanding of what AI actually does well.
What AI Changed About Coding
Let me be honest about what AI does in my daily work. I use AI coding assistants extensively. They autocomplete functions, generate boilerplate, write tests, and help me explore unfamiliar codebases. They make me faster. Significantly faster.
But here’s what they don’t do: they don’t decide what to build. They don’t understand why a client needs a specific feature. They don’t architect systems that balance competing constraints — performance, cost, maintainability, security, user experience. They don’t debug a production outage at 2 AM by reasoning about system behavior from incomplete logs.
AI changed coding from “translating logic into syntax” to “directing and evaluating machine-generated solutions.” The syntax part — the part that most “learn to code” courses focus on — is exactly the part that matters least now.
This is actually good news for the next generation. The tedious part of programming is going away. The interesting part — the thinking — is more valuable than ever.
Skills That Survive AI
When I think about what I want my kids to learn, I don’t think in terms of programming languages. I think about cognitive skills that AI amplifies rather than replaces.
Problem Decomposition
The ability to take a complex, messy, real-world problem and break it into smaller, solvable pieces. This is the core skill of engineering, and it’s the core skill of effective AI use.
When someone says “we need to automate our invoice processing,” the skill isn’t knowing Python. It’s asking the right questions: What format do invoices arrive in? What data needs to be extracted? Where does it need to go? What happens when extraction fails? What’s the approval workflow? How do we handle exceptions?
Breaking big problems into small, clear steps is how you direct AI effectively. It’s also how you solve problems without AI. It’s a universal skill.
Systems Thinking
Understanding how parts connect to form wholes. Why changing one component affects others. How feedback loops work. Where bottlenecks form.
AI is great at optimizing individual components. Humans are needed to understand the system — to see that making the checkout flow faster won’t help if the inventory system can’t keep up, or that automating email responses is useless if the underlying customer problem isn’t being solved.
Kids who understand systems — whether they learn it through building with LEGO, designing Minecraft redstone circuits, or yes, writing programs — will have an advantage in any field.
Communication
The ability to explain technical concepts to non-technical people. To listen to a stakeholder’s problem and translate it into a technical specification. To write documentation that humans can actually understand.
AI struggles with context, nuance, and the political realities of organizations. The engineer who can explain to a CEO why the “simple” feature request will take three months — and propose an alternative that achieves the business goal in three weeks — will always be valuable.
Critical Evaluation
When AI generates a solution, someone needs to evaluate whether it’s correct, efficient, secure, and appropriate. This requires understanding not just what the code does, but what it should do. Kids need to develop judgment — the ability to look at an answer and ask “is this right?” rather than accepting the first output.
This is especially important because AI is confidently wrong with alarming regularity. The skill of questioning plausible-sounding answers will only become more important.
What I’m Teaching My Own Kids
I’m not teaching my kids Python syntax. Here’s what I am doing:
Logic puzzles and games. Chess, strategy games, mathematical puzzles. These build the pattern recognition and logical reasoning that underpin all technical thinking. More importantly, they’re fun — my kids actually want to do them.
Building things. Physical and digital. LEGO Technic for mechanical thinking. Scratch and similar visual programming for computational thinking. The goal isn’t to produce code — it’s to experience the cycle of imagining something, planning how to build it, encountering problems, and iterating until it works.
Asking “how does this work?” When we use an app, I ask them how they think it works. When something breaks, we talk about why. Curiosity about the systems we use every day is a habit that compounds over time.
Using AI tools together. My kids see me use AI assistants. We talk about what the AI does well and where it makes mistakes. They’re learning to treat AI as a tool — useful but not infallible — rather than as magic or as a threat.
Writing and explaining. I encourage them to explain things to each other, to write down their ideas, to present their projects. Clear communication is the meta-skill that makes every other skill more effective.
The Real Answer
Should your kids learn to code? Yes — but not the way most people mean it.
Don’t enroll them in a course that teaches JavaScript syntax for six months. That knowledge has a shelf life measured in years, not decades. The syntax your kids learn today might be irrelevant by the time they enter the workforce.
Instead, help them develop the thinking skills that coding teaches at its best: breaking down problems, thinking in systems, evaluating solutions, and communicating clearly. These skills are language-agnostic and AI-proof.
If your kids enjoy coding, let them code. If they prefer building robots, making music, designing games, or writing stories — those activities develop the same underlying skills. The medium matters less than the mindset.
The kids who will thrive aren’t the ones who can write the most code. They’re the ones who can think clearly about problems, direct AI tools effectively, and communicate their ideas to other humans. Teach them that, and the specific tools they use will take care of themselves.
One More Thing
If you’re in tech yourself and wondering how AI will change your career trajectory, or if you’re a business leader trying to understand what technical skills to hire for, these are conversations I have regularly. Drop me a message — I’m always happy to share my perspective.