Introduction#

I’ve started coding when I was about 12 years old struggling to put together Minecraft plugins written in Java.

I certainly wasn’t good at it but I enjoyed seeing my creations in game, changing my (and my friends) experience and most importantly: the process of getting there. It was more about discovering how computers work and over a couple of (admittedly not so great) discord bots it all lead to me now studying CS at a german uni for applied sciences.

In my lectures there are a lot of guys who never did anything in terms of programming, networking etc before and as such had no prior experience. Of course that isn’t required to get good at it at all, but it certainly helps a lot. I think most of the people who really have a deep desire for knowledge about CS stuff have already made prior experience in that field just because how easy it is to fire up google and search “how do i get started with programming?”, “how to build a website” or even “how to hack into someones pc”.

All of the people who have that prior experience I am talking about have one thing in common: They learned it the hard way. They spent a lot of time reading, trying new stuff and recompiling the same project with no changes because they had no idea how the toolchain, the compiler or even the language they were using worked.

What do you enjoy about programming?#

That process “filtered” out the people who didn’t really enjoy doing programming which is of course completly fine, but it forced you to figure out what you enjoy about it and whether you really enjoy the topic at all. At first you simply copy paste stuff from Stackoverflow but as the time goes on you start to understand what the code you’re copying really does. You either have to choice to stop and continue on with your life, or go down the rabbithole and understand more fundamental things every day.

Everyone probably had that moment where they simply stopped their current side project because of some sort of blocker they couldn’t come over. But then after a few weeks/months one day you think of trying it again. It’s similar to am addiction I guess. I love being able to solve problems in my daily life and make it just that little bit easier on me.

The Problem#

I have discovered that most people getting started today are using ChatGPT, Github Copilot, Claude or any other of the countless coding assistants. I do so as well.

I believe to have observed a trend of people not questioning the code they write (same thing as with pasting from Stackoverflow). However in this case there is a big difference: Beginners are never forced to think about what they write, simply because their AI assistant understands it and is able to change it precisely to their needs. That way you never start to run into that roadblock where you can’t continue your project because noone before had that specific problem you’re facing (or its to obvious to have been posted onto a online forum).

Don’t get me wrong: I use assistants as well (even if rarely). There is certainly a place for them to exist in the developer community, even simply because there obviously is demand for them. Personally I think not using them helps keeping your brain trained and having complete control over what you do.

That “forgetting” is especially dangerous if you have never even had the knowledge in the first place and while trying to aquire it simultaneously getting everything “spoonfed” by an AI assistant. My theory is that that will consequently lead to you never outgrowing the junior dev status because thats what the assistants are. They help you do simple tasks, but in cases where they fail (which are becoming less common, but they will always still be there) you are on your own. If you now never made the experience of searching stuff up on your own, thinking about projecting the solution to other peoples problems onto yours, etc you will be stuck.

I totally understand the temptation of just going with the easy option, asking ChatGPT to implement a simple algorithm instead of thinking about it. In my mind thinking about the problem and trying to solve it yourself results in a much better code which you actually are able to fix if it breaks.

In the end these models are only trained on the stuff you are working with anyways. Every company trains on the openly accessible data on the internet. Everything these assistants know you can know as well by searching on google.

The thing that differentiates a junior programmer from a senior one (and for that matter a good from a great one) is not the amount of knowledge they have access to. That is the same for everyone. Its the amount of systems knowledge you have, how the different subsystems affect each other and work together.

In “employing” AI to do the problem solving for you, you are relying on a third party to do the thinking for you. And if someone else does the thinking, what is your contribution to the project?

The Solution#

This goes out to all the people wanting to start out programming, beginning their first semester of CS at a university or everyone else just curious about the topic:

Please, PLEASE do yourself a favor and try to conciously abstain from using those assistant (for the most part). You are outsourcing the thinking and systems knowledge to a third party. They are an incredible tool and no doubt help you with simple tasks. But thats it. They can’t replace your brain.

You will become a much better programmer if you choose to challenge your brain, start to understand read documentation. Thats probably the most important thing: You need to start reading the documentation of the frameworks your working with. Thats the only way to get into its philosopy, inner workings, whatever you call it.

But in the end, its your choice.

Do you choose convenience or curiosity?