During the summer, I had a few days off as summer vacations and, besides getting some much-needed rest, I decided experiment and to see if I could build an application using nothing but artificial intelligence. So, I picked up a project I had wanted to build for quite some time and got started.
The project was a web application for predicting EuroJackpot numbers based on historical data, along with the presentation of various statistics. After setting up a Laravel application, I started prompting. The project was finished in about an hour. Since I had finished much faster, almost at the speed of light, than I had expected, I decided to try using AI to make some modifications to another React project as well. I completed those changes in a few minutes too. So, something that would have taken me several hours was finished in about an hour. And that’s where the positives ended.
The fundamental problem with all of this was that this one hour gave me the illusion that I had actually accomplished something without really using my brain, without using my creativity, and without putting in any real effort. I simply described what I wanted to do and the magic genie did it for me. It gave me the same feeling someone might have when they hit a jackpot on a slot machine. My brain remained inactive, while the addiction to “increased productivity” kept growing.
I had built the React project a year and a half ago, and I no longer remembered exactly what I had done. I didn’t remember the components, I didn’t remember the routes, and if I hadn’t used AI, I would have needed some time to read through what I had done, bring it back into my memory, and then move on to making the changes. The fact that I didn’t have to do that is, in my eyes, extremely negative. And that’s because if I need to make a larger change right now, I don’t remember anything. With the new additions, which were made by AI, the complexity of the project increased without me being actively involved. So even if I sit down and read through what I had built a year and a half ago, I will now also have to analyze changes that I never actually saw or experienced myself.
If I had to summarize the whole experience in a single word, it would be: illusion.
If I didn’t have the self-awareness to recognize what was happening, the LLM would have left me with the satisfying feeling that I had used my creativity, that I had learned things, that I had accomplished something important, when in reality I had simply written a few descriptions.
I understand that speed is important. But are we ultimately going to sacrifice our knowledge and creativity for more output? And are we really going to rely on a machine that, on the one hand, does not produce deterministic results like, for example, an assembly line in a factory, and on the other hand, ties us even more closely to huge companies that sell us tokens? For years, I had learned that I shouldn’t tightly couple my code to a particular technology. Yet now, the exact opposite seems to be happening.
Even after this more complete experience I had with AI, I still consider it a tool. I don’t want it to replace my creativity or the process we have followed for so many years to gain better experience in software engineering.
Because the truth is that, right now, the only reason AI actually works professionally in software development is quite clearly because there are still experienced programmers who have fallen flat on their faces countless times in the past while writing code by hand.