The Learning Loop and LLMs

Original post: The Learning Loop and LLMs

Last week, I came across this disclaimer on a Github project. The author explains how they vibe coded the most part of it.

This sentence struck me (emphasis mine) as it reflects the exact way I’m feeling when I rely too much on LLMs:

Overall I would say it was a horrible experience, even though it took 10 hours to write close to 10000 lines of code, I don’t consider this my project, and I have no sense of accomplishment or growth.

While vibe coding we all have that sense of instability for not having the full code context in our minds, not knowing if the code is correct or not. And this gets worse if the language we’re coding in is not one we’re familiar with.

I also found a similar sentiment in a previous post I shared.

So it seems that yes, tools are useful, but the more we use them, the more we lose enthusiasm in what we do.

Today, I’m cross-posting an article by Unmesh Joshi. In my mind, it actually closes the loop and makes this feeling very explicit. The culprit of everything lies in the missing learning experience:

An AI can generate a perfect solution in seconds, but it cannot give you the experience you gain from the struggle of creating it yourself. The small failures and the “aha!” moments are essential features of learning

As software engineers, learning is a fundamental part of our job. That’s what makes our profession unique and fulfilling. Don’t let it die out because:

Our ability to respond to change comes not from how fast we can produce code, but from how deeply we understand the system we are shaping.