Born into the tool

There is a particular kind of knowledge that only arrives with time. It is not the knowledge you can look up. It is the kind that settles into you slowly, through repetition, error and the quiet satisfaction of finally understanding why something works. I have been thinking about that kind of knowledge a great deal lately, because I suspect it is about to become both more valuable and much harder to acquire.

This is partly a story about how I learnt my trade. It is mostly a question about how the next generation will learn theirs.

Learning one layer at a time

I did not begin as a developer. I began with networks: cables, switches, addresses, and the stubborn logic of machines that either talked to each other or did not. Networking taught me something early that I have never lost. Every system rests on layers, and when something breaks, the fault is usually one layer below where you are looking.

Coding came next, almost as a consequence. Once you understand how machines exchange information, you start wanting to tell them what to say. I learnt by reading, by copying, by breaking things and then sitting with them until they worked again. Much of that time was frustrating. Most of the progress was invisible while it was happening.

Then the web arrived, and with it a whole new set of things I did not know. Suddenly it was not enough for something to work. It had to be understood by a person on the other side of a screen. So I learnt design, not because anyone asked me to, but because I could see that a working system nobody could use was not really working at all.

Looking back, what drove all of it was not ambition. It was curiosity, the kind that does not switch off. A hunger to know how the next layer worked, and the one beneath that. There was struggle in every stage: long nights, manuals that assumed knowledge I did not have, problems that took days to solve and turned out to be a single misplaced character. The struggle was not a detour from the learning. It was the learning.

Back where I started

Something interesting has happened in the last few years. With enormous bodies of knowledge now compressed into models, and assistants like ChatGPT, Claude, Grok and Copilot sitting beside us as working colleagues, I feel as though I am back in the era I was born into professionally. Everything is new again. The rules are being written as we go. Curiosity is once more the most useful skill in the room.

But the role has shifted. When I started, I was the one doing the building, line by line. Now the building can be done with me, and often faster than I could do it alone. What remains distinctly human is the vision: deciding what is worth making, imagining what does not yet exist, and holding the shape of an idea while the details are produced around it. We have become the ideators, the innovators, the people who set direction.

That would be a comfortable place to stop, except for one thing. These new colleagues make mistakes. They make them confidently and fluently, in perfectly formatted prose and code that looks entirely plausible. They drift. They need steering.

The only reason I can steer them is the body of knowledge I built the slow way. When an assistant proposes an architecture, I can sense where it will fail because I have watched similar systems fail. When it writes code, I can see the subtle error because I once spent a weekend chasing the same one. When it confidently states something about a network, I know which layer it has misunderstood. My judgement is not magic. It is sediment, laid down over decades of learning things the hard way.

A tool can extend what you know. It cannot supply the ground you stand on to judge it.

A generation born into the tool

This is where my wonder turns into a genuine question.

We have already watched one generation grow up inside social media, born into the feed rather than arriving at it. Many of them navigate it with remarkable fluency. Yet fluency in a tool is not the same as understanding what the tool is doing to you. Telling a true story from a persuasive one, or a healthy community from a corrosive one, requires something the feed itself does not teach.

Now a new generation is being born into AI. They will never know a time when an answer was not instantly available, or when a first draft required a blank page and an hour of struggle. In many ways that is a gift. The barriers I spent years climbing will simply not be there.

But I wonder what fills the space where the struggle used to be. If the body of knowledge is never built, I am not sure what grounds the mind. It is hard to recognise a wrong answer if you have never had to work out a right one, and harder still to tell good from harmful, or true from merely convincing, when every judgement can be outsourced before it is formed.

I do not think the answer is to withhold the tools. That would be neither possible nor wise. The tools are extraordinary, and refusing them would be its own kind of failure. Perhaps the answer is to be deliberate about what we still ask people to learn by hand: the fundamentals, the layers beneath, the patient work of understanding why. Not because the machine cannot do it, but because a person who has done it carries something the machine cannot give them.

Curiosity was always the engine. Struggle was the forge. The tools have changed, and they will keep changing. What I hope does not change is the willingness to go one layer deeper than the answer you were handed, and to keep doing that long enough for it to become who you are.