Where Does It All Start to Go Down

Lately, with the convenience of LLM in my humble daily life as a student studying Computer Science in some backwater university in Indonesia. I have met with the very idea of not having to lift a single finger to complete an assignment by relying on LLM, namely, the all-time popular, everyone's favorite, and infamous ChatGPT. It was released in November 22t, 2022, and gain traction quickly, with Indonesia being no exception.

I still recall the first time I used it I felt like having a personal holy grail to solve all kinds of my college assignments without having to think twice about it. Being a freshman, most of my subjects are fundamentals--arguably so easy that I didn't need AI's help to learn them. But knowing how easy they 'hypothetically' should be made me think, "Uh, these are some real easy-peasy thingies. I'll just breeze through them with my personal 'holy grail' for now. I'll manage some time to study it later when I need it." Little did I know that in the future this would come back bite me in the ass, and the future is now, and my ass hurts so bad.

"Eh.. Easy, I'll Just Had to Do it This Way, Do that That Way, and--Wait.. What The Fuck?"

It was a Tuesday morning, I attended the first session of my Data Structures class in my second semester. Being the clueless, overconfident fuck that I was, I just slept through the whole lecture and didn't pay any single attention. "Eh, I'll just learn this with ChatGPT later in the afternoon," I thought. Well, unsurprisingly, I didn't manage to do that. And just like that, my tuesday routine for the entire semester is set in stone. How did it go, You ask? What? Linked Lists? What are those? Wait--lemme ask ChatGPT about that real quick.

The ChatGPT Shortcut: A Double-Edged Sword

The first time I tried to look up Linked Lists on ChatGPT, it felt like magic. I typed in a quick query, got back a perfectly structured explanation, and even some code examples. For a moment, I thought, "Wow, this is amazing! Why would anyone actually go through the trouble of learning this when ChatGPT can just summarize it for me?"

But then came the twist. ChatGPT, in all its glory, has a limited context and sometimes an odd sense of priority when answering. I mean, how the hell did I end up getting a full explanation of the concept of Linked Lists without knowing what a node is in the first place? Or how about that algorithm where I needed a good grasp on recursion but instead got a vague description of a "for loop"? Was this the power of LLMs, or was it a curse in disguise?

The more I relied on it, the more I realized that it didn't help me learn the foundational concepts needed to tackle the bigger picture. All the shortcuts I took ended up adding more confusion than clarity, and it wasn't until much later, when things started to feel like a blur of fragmented concepts, that I understood how screwed I really was.

The Stomach-Churning Revelation

Nope, the revelation didn't hit me during some dramatic midterm panic. Not even close. It hit me much, much later—sixth semester, to be exact. I was doing this pretty final project at Bangkit, the kind of project you actually wanna show off on your LinkedIn or portfolio or whatever. It had to be the project. Clean code, solid backend, good design, all that jazz.

I was designing the Firestore database schema. At first, things looked fine. Until I had to store a list of dynamic, ordered data. Like, data that can be inserted in between, deleted from the middle, reordered without burning everything to the ground. And guess what I did?

I used a damn array.

An array.

Like a full-on pleb who never heard of Linked Lists or node pointers or any of that. All those Data Structures lectures I skipped, all the times I told myself "I'll learn it later with ChatGPT," it all came back and smacked me across the face. Right then and there, while I was wrestling with Firestore's limitations, trying to hack my way around missing indices and shuffling array elements manually like a caveman with a stick.

And all I could think was: "If only I truly understood how Linked Lists work, I wouldn't be stuck with this janky-ass workaround."

The worst part? It wasn't even about just understanding the concept. It was the fact that I never practiced it. Never internalized it. Never gave it the respect it deserved because, well, ChatGPT was always there to do the thinking for me. I let it think for me, not with me.

And so, there I was. Sixth semester, allegedly a "final project" tier student, still stuck in the same pitfall I dug myself into way back in second semester. Only this time, the pit was deeper, and there was no LLM rope long enough to pull me out.

Now What?

Looking back, I can't help but laugh at how blindly I trusted shortcuts to carry me through my academic journey. I'd used ChatGPT like it was my personal cheat sheet, thinking I was outsmarting the system. But in reality, I was just setting myself up for failure in the long run. Sure, it worked for a while—when assignments were small, and exams were simple. But once the projects got bigger, the stakes higher, the holes in my understanding started to show.

The project at Bangkit was the moment it all clicked (a bit too late, I might add). I realized that knowledge is not something you can just rent for a quick fix when you're in a pinch. It needs to be built, tested, and most importantly, practiced. I was left scrambling to patch up my weak foundations, wishing I had put in the work earlier.

I also realized something else: learning isn't about being perfect, it's about doing the work. The act of struggling through a problem, even if it feels frustrating as hell, is what truly teaches you. That's the stuff that sticks. The stuff that helps you build real solutions, not just copy-paste ones.

From that point on, I made a promise to myself to not rely on shortcuts anymore—not unless I really understood why it worked, how it worked, and when to use it properly. I've started putting in the effort to understand things at a deeper level, even if it means stepping outside my comfort zone and doing the hard work. It's the only way to avoid becoming that guy who's just a walking pile of AI-generated answers but can't actually solve problems in the real world.

Yeah, maybe Linked Lists weren't the sexiest thing to learn, but they were a lesson I needed. Firestore's schema design, and any other complex project I'll tackle in the future, will be better because I decided to stop being lazy and learn the core concepts. I'll be better for it, and hopefully, that will show in the projects I'll take on in the future.

So, if you're reading this and you're tempted to take the shortcut with LLMs, just remember: they can't fill in the gaps in your knowledge. At the end of the day, you'll still have to do the hard part—learning, applying, and failing until you get it right.

And trust me, doing it the right way is far more rewarding than just breezing through it.