Okay, I was wrong

Last week I thought to myself: “Well, I have a few tasks. Most of them are just simple refactorings, so I shouldn’t spend too much time thinking beyond the scope of my tasks.” They were relatively well-written, and I only needed to make a few changes to adjust them as necessary. This was on Thursday. I spent another 10 hours on Friday, and on Saturday morning, I still had to work. But before wasting another 10 hours, within the first 30 minutes, I realized that nothing would get resolved there.

Given that, I managed in 4 hours what hadn’t been done in 14. Putting it in perspective, my problem isn’t AI itself, but me. Honestly, certain questions come and go. For relatively simple tasks, you won’t spend more than 20 to 30 minutes solving something with AI, maybe 10 to 20 minutes more without it. However, for more complex, abstract, older, or poorly documented tasks, AI becomes less useful, and the same questions arise again.

• Is it worth it?
• What would truly be worthwhile?
• Does this represent a real productivity gain?
• Am I actually exercising my ability to understand what’s being done?
• Am I spending unnecessary effort trying to understand something when I could just do it?

I know some of these questions are even the subject of research in universities, such as the MIT study (https://arxiv.org/pdf/2506.08872v1). Another point is that, given our idle capacity, we may end up spending time on useless things. Honestly, much of the time many of us spend isn’t for study. In a dopamine-driven, unfocused world, it is very easy to shift attention to other things after opening a new tab.

I’m still reflecting on the pros and cons. To be honest again, I use AI constantly and experimentally, but at times it makes me waste too much time in a limbo of uncertainty. Is it worth spending so much time to save something that might not actually be saving anything? Not to mention the financial cost. It is easy to spend a significant amount on something we often use merely for convenience, with a false sense of productivity.

One last point: I miss my nvim. I like the cursor and other editors, but more and more everything feels too easy, too magical, and that bothers me. It makes me feel less capable, perhaps even ignorant, and adds to the trade-offs I need to consider. I might need to rethink some tools and better organize my doubts as a programmer. I need to tackle things point by point until I reach final and competent answers.

Anyway, that’s it for today. AWS went down again, and I’ll take the time to better organize my thoughts.