There will be a day when a new model breaks every record and becomes so powerful that my whole career of acquiring knowledge simply vanishes. It will make action far more rewarding than it is now. Amid the reasonable worries about this technology, I’ll be grateful, because it took me out of a group I never really belonged to, just like when I was 13.
Reading Hacker News, I often didn’t realize just how much slop I might be producing throughout the day as a software developer.
When I was 13, I started my lifelong journey as software developer, and that kind of ends today. I programmed purely out of joy, as many of you did, because I didn’t have to work, and fell in love with everything I could make from my home.
Given my pure joy for creating new things, I eventually found a profession complex and deep enough to keep me engaged, one I could pursue entirely on my own, and broad enough to let me explore so many different areas, because I simply love creating things, until something happened.
What I loved was the freedom to make whatever I imagined, and the excitement of seeing something of mine working on a screen, again and again.
When LLMs became good enough at coding, I went crazy thinking about all the possibilities. I could now build things faster than ever. As developers, we were privileged because we already knew how to code, while non-technical people still struggled to build things with LLMs. Even a little programming knowledge made a huge difference in what you could create and how fast you could do it.
But that advantage was about to disappear as the knowledge needed to build things became less and less necessary. Soon, everyone would be cooking up the same things I could. This had been slowly happening over the last few years…
Most of my attention always went to the final stage, leaving the process out. Frameworks and techniques were never that important to me. That is why I got addicted to this drug differently from other software developers: I enjoyed the destination far more than the journey. Now that the journey doesn’t exist from my perspective, I realized it lost a little of its charm. Part of it had to do with the challenge itself. I still think the result is what matters most. Before AI, effort was scarce, so we thought carefully about what was worth building. Now that the effort has shrunk, we don’t identify with it anymore, and creating feels a bit boring. The same happens with any abundance. Our days are full of small miracles that become normal, and I try to stay grateful for them.
I would classify most of the people I interact with into two groups. I have developer friends who are addicted to their LLM tools, crazy setups, and keep building things down to their bones with good performance and quality, more or less, and I know businesspeople who are addicted to the idea of building things without needing developers, with the same distribution of quality. Both groups are obsessed with being able to build.
Trying to get out of the first group today, I’ve become some sort of technical thinker who mostly types instructions and makes decisions seeing a terminal. Most of my day is approving and reviewing. I started using my brain less, yet somehow kept trying to do more of the same. More of what everyone else is already doing. Am I becoming bored with this world? The grass on the other side looks much greener. Why would I keep programming as my main focus when I could be making videos, painting, or starting so many other side quests in life that give me something software no longer does?
I tried, in that sense, to stay away from the drug of fast creation, spending less time programming, and focus on other things. I feel less productive, but that bothers me less than being ungrateful for technology I always dreamed of having access to. I’m at a point in my life where I want to prompt less, trusting that the next model will be better. And so far, that has always been the case… So I want to put my attention into making videos, painting, writing and a few side quests that don’t involve software.
It built up slowly until I stopped being able to ignore it. The kid I was at 13 would have called these tools magic, and I’m the one setting them aside.
I’m just an average developer, and when the time comes and all those main LLM benchmarks are saturated by that all-powerful model, with who knows what kind of new physical techniques yet to be discovered, no trace will remain of that decade of programming experience. I don’t need that day to arrive to leave. I will be left with only muscle-memory vestiges of a time when I could type commands to a machine directly from my thoughts, without any kind of intermediary, and, trying to see the positive side, I will be grateful for having split my life in two, for launching myself into a new life where there are so many more things to do than prompt websites and apps.