A few weeks ago a young boy who had just started university came to me for advice.

"I want to do what you do. At uni we are learning Java, but I'm not sure if it's the right thing to learn?"

Two years ago I would have said "learn JavaScript" without thinking twice. This time I froze. So much had changed that I was worried my advice might be outdated within a year or two, so I looked at him and said, "Maybe learn AI."

The sceptic

I love technology and I love building, so when AI arrived I was all over it. I made fun of ChatGPT for not knowing how many Rs are in "strawberry", and then I watched my parents' faces when I took a photo of the table and the AI named every single item on it. Amazing and scary at the same time.

When people started saying programmers were finished I laughed, because Copilot could barely autocomplete and every week there was another story about an AI deleting someone's database. I am more reliable than that, I won't be replaced. And then, faster than I would have imagined, the models became bigger, better, smarter, faster and cheaper, and before I noticed I was giving the AI more and more work and trusting it with more complex problems.

My curiosity took over. I spent my nights and mornings reading articles and X posts about which model had just come out, which skills people were using and what the latest harness was, and during the day I tested it all to form my own opinion. Some of it was great and some of it I hated. At the end of the week I shared what I had picked with the team, and a few colleagues adopted it while others probably thought I was crazy. I was never sure which tools were really worth sharing and most days I felt overwhelmed by all the shiny new things, but it felt good to be learning and helping my friends at the same time. But at what cost?

What I lost

I love the craft. I loved finding clever ways to solve hard problems, spending three days stuck and waking up one morning with the answer, or cutting the load time of an app. I even had a silly motto, "Everything is possible", because whatever the task there was always a way and I loved the challenge. Then AI started doing it, and I told myself it was fine, the AI is faster and now I can deliver more value, it was a race; we could not create tickets as fast as I closed them. But something was missing. The code had become strange to me, I had lost the connection with it, and it did not feel like mine anymore.

My father once told me, "If you do something you love, you won't work a day in your life", and I always felt lucky to work in a field I love. Now it was slipping away and I lied to myself about it: it's a phase, I'm just tired, probably because last night I stayed up late to give the LLM one more prompt so it could run while I slept, and I have to be productive anyway. The truth is I had started to hate it. Walls of AI text where I had to dig for one or two good ideas, design docs of four to six pages, fully generated and written in pure claudish. And then a message from a colleague that started with "You can reply to Razvan that he can do...", and I was speechless, this person had not spent two minutes reading or editing a three line answer before sending it to me.

AI was making me hate the thing I used to love, and I felt lost.

I always enjoyed sharing cool tools and having interesting debates about technology, but now every post I shared and every document I wrote for feedback got ignored. Oh no wait, someone reacted with a 👍. Suddenly everyone was busy talking to their own AI, and nobody was available to give feedback or spend thirty minutes debating how to improve that slow feature.

Looking for an exit

I started questioning everything. Maybe the people who said AI would take my job were right, just not the way they meant it. I was not going to be fired and replaced by three Claude agents, I had simply lost my passion, so what else could I do? I like making things. I pictured a little workshop with 3D printers and a CNC machine, and maybe I would build fishing lures, why not. In uni I built mini sumo robots so how hard could it be?

Or maybe I should reduce the time I spend online and find some other hobbies. Go to the gym? Start running? Maybe I'm just not that young anymore and can't spend twelve hours a day at a laptop. Then I remembered that during covid I had started learning Chinese, so I could go back to that, and now I had AI, which is so smart that it would surely help. I tried ChatGPT and a handful of other AI powered products and they were all bad, nothing clicked. But wait, I can code. I can make a better app than these random people on the internet and save some money too. So I opened Claude Code and started building my own language app.

Three months later it was still under construction. I had worked on it every weekend and had not learned a single new word. Sigh. The build had captured me so completely that I never found time to practise. I had an app that was kinda working, but not really, and it did not help me at all so what was the point in doing it. I had to start from scratch three times because this stupid AI kept breaking everything (I even specified “no mistakes”). The whole process felt frustrating again. I was in the same place where I began, building something that was not mine, something I would not care about if I deleted the code.

The walk

Once again I felt lost, so I went for a walk.

I stopped at the first shop, bought chips and a bottle of juice, and set off. It was a summer day, 37 degrees, and I walked slowly through rural Chinese streets along the Yangtze river. It was calm, and so far from the noise in my head. 3 hours and 12km gave me enough time to think. I realised that something had already happened without me noticing. I had spent so much time building guardrails to stop the AI from making mistakes that I had learned its weak spots and its strengths. I had built a bunch of small tools to organise my day to day life, from scripts to skills to complete workflows where my agents ran for hours without supervision and without me worrying about what they would break. I had changed. I had adapted. That was the moment I realised I was never in love with writing the code. It was learning and building, it was always the New. But the New is also scary and uncomfortable sometimes.

What changed

For years I thought clever solutions and beautiful implementations were all I was good at. But now I had time to use the product I worked on every day and I felt its pain, and changing it was easy, I just had to ask Claude. So I built prototypes, little UIs I could click through to feel how the product would behave, and then had the agent build the one that felt right. A year earlier I would have shrugged and said, "Users will get used to it, it is a matter of teaching them, change is always hard." Now I could not do that. I spent a few extra days polishing until it was easy and smooth to use, and I was proud again.

The code may not be mine any more, but the result is. It comes from my effort to deliver a high quality experience. No user will ever care how readable my code is or how well I formatted my comments, they care whether they can do their job quickly without fighting against the app. The same joy came from the other part of my week, when I spent hours showing colleagues my setup and the tools I use, and watching them get faster felt as good as my best optimisations ever did.

It finally clicked in a 1-1 with my manager. I was describing how much more I could build now, and the fact that I don’t need to worry about the code implementation, allowed me to care more about user experience. While I was saying it I realised that writing code was never the job. It was the language we knew for talking to the computer. The job was always to puzzle together clever solutions, make hard and informed decisions, and really understand the problem we are solving for the user.

Back to the kid

Would I still tell him to learn AI? Yes, but now I know what I mean.

A few friends asked me which tutorial to watch to learn this AI thing, and I don’t think there is one. You cannot learn this from a book or a YouTube video. It is changing too fast. Learning AI means being curious: try it, push the limits, ask stupid questions, give it impossible tasks. Sometimes it will surprise you and sometimes you will laugh at how dumb it is, and once you know where it is strong and where it is weak, you start to see where you can use it. That is what happened to me. I stayed curious and kept pushing to see what came next, and that curiosity led me back to what I loved.

AI is a tool, make sure to learn how to use it to your advantage. Use it to learn, use it to explore, but always be curious. So, to the boy: Yes! Learn AI.