It’s 11 PM on a Tuesday in late 2025. I have a Data Structures assignment due tomorrow. The task? Implement a Red-Black Tree from scratch in C++.
My professor, bless their old-school heart, wants us to "feel the pain of pointer management."
My brain, fried from three lectures today, is screaming for an easier way. I open my IDE. I have GitHub Copilot Workspace open on one monitor, and Claude 3.5 Sonnet on the other.
I type: "Yo, I need a Red-Black Tree implementation in C++. Keep the vibe academic, handle the edge cases for rotation, and add comments that sound like a stressed undergrad wrote them."
Thirty seconds later. Done. ✨
It compiles. It passes the test cases. It’s beautiful code. And I have absolutely no idea how it works.
This is "Vibe Coding." It’s the dominant strategy for half my cohort, and honestly? I think it might be ruining my education.
The Sugar Rush of Speed 🍭
"Vibe Coding" is the evolution of prompt engineering. We aren't writing detailed specs anymore. We're just describing the intent—the vibe—and letting the models handle the boilerplate, the syntax, and the logic.
It feels incredible. It’s intoxicating. What used to take me a weekend of agonizing over Stack Overflow threads now takes an hour of casual conversation with an LLM. I’m shipping side projects faster than ever. My GitHub contribution graph looks insane.
I feel productive. I feel like a 10x developer.
But then I sit in my Operating Systems lecture, and the professor asks a question about memory paging or race conditions, and my mind goes blank. The AI never mentioned that. The AI just made the "Segmentation Fault" go away so I could go back to scrolling TikTok.
Imposter Syndrome 2.0
🎭
Every CS student has imposter syndrome. But in 2025, it’s different. It’s not just "I’m not smart enough." It’s "I’m a fraud who is just really good at gluing together AI outputs."
The terrifying realization hit me during a mock technical interview last month. I wasn't allowed to use Copilot. I had to reverse a linked list on a whiteboard.
I froze.
I knew what a linked list was. I knew why you’d reverse it. But the actual muscle memory of moving the pointers? The algorithm? It was gone. Atrophied because I’d outsourced that kind of "low-level thinking" to the machines for the last two years.
It felt like I was a carpenter who only knew how to order pre-built cabinets from IKEA, suddenly asked to build one from raw lumber.
The CS Degree vs. The Real World
This creates a massive conflict in university right now.
The Academia Vibe: "You must understand the fundamentals! Build it in C! Understand the Big O notation!"
The Industry Vibe (in 2025): "Ship it yesterday. Use whatever tools make you fastest. If you aren't using AI, you're already behind."
If I keep Vibe Coding my way through this degree, am I going to be useless when the AI hits a wall it can't climb?
Finding the Balance: From "Do It" to "Teach Me"
I’m not going to stop using AI. That would be professional suicide in 2025. Pandora’s box is open.
But I am changing how I use it. I’m trying to move from passive consumption to active interrogation.
If I Vibe Code a solution, I force myself to do what I call a "Reverse Vibe Check":
Don't just copy-paste: I type out the AI-generated code myself. It sounds stupid, but it forces my brain to process the syntax.
The "Explain Like I'm 5" Prompt: Once the code works, I open a fresh chat and paste it back in: "Explain this code to me like I’m a first-year CS student who hates math. Focus on the 'why', not just the 'what'."
Break it on purpose: I ask the AI, "What are three ways this implementation could fail in a high-load production environment?" This teaches me the edge cases the happy-path "vibe" missed.
The Verdict?
Vibe Coding isn't ruining my degree. My reliance on it to skip the hard parts is ruining my degree.
AI is the best tutor ever invented, but it's also the world's easiest enabler of laziness. The challenge for us juniors in late 2025 isn't learning to prompt better; it's having the discipline to close the chat window and stare at the broken code until we understand why it’s broken.
Are you Team Vibe😎🍹 or Team Grind? 😢🧠
Top comments (63)
"Team vibe" is straight way to unemployment.
In a few years it is not you who will be vibe coding. Your Product Owner / Product Designer will. They can use keyboard too to write prompt with feature request. You will be an obsolete layer in this process.
The demand for developers will be reduced to hardcore stuff. Backbone systems, mainframes, very complex backends, device drivers, bare metal code, etc. You will be paid to understand complex problems.
This is assuming the current model is sustainable. With the AI bubble continuing to grow yet nobody finding a way to make money from it, while the internet is getting saturated with already AI generated content and the people generating the quality training data are getting fed up with their work being stolen and starting to take measures against it, I wouldn't be too surprised if AI crashed really hard in the coming years and only made a full comeback after many years of hardware breakthroughs make it financially viable again.
It is sustainable. Or to be more precise - it reached a level which makes it reasonably priced tool for software development.
AI bubble will crash hard. All AI FOMO and AI Slop projects will be evaluated by the market, which is natural healing process of oversaturated economy. But AI development tools will survive this crash for three reasons:
This faces the same feedback problem as most other AI training: Real people start moving their code out of reach from AI tools and more and more of what remains is itself AI-generated. Eventually the quality just degrades.
And in the process make mistakes more likely.
Compilers don't catch difficult mistakes. If your standard for good AI code is that it compiles, then yea, compilers make that easy; but for one they aren't "instant", specially on bigger projects, and more importantly, they don't help much with the real problem of AI code.
But all of that aside, my point still stands. AI doesn't pay for itself; it's not financially sustainable. And that's likely going to get worse as the massive resource consumption brings AI into conflict with local populations around server farms.
The bubble will burst eventually, AI will "die" for a while, and eventually, technology will have caught up to its requirements and companies will start to scale up its usage again, probably with much less hype surrounding it.
You’re absolutely right
Vibe coding assignments is like copying your friends homework, but your friend is the tip of your fingers where you can just ask for the answer. I don't blame students for going down this route at all. When the stress and pressure builds up, it's hard not to make these choices in fear of failing out.
Problem is, now every student will blend together with all the cheating. So it really comes down to your side projects to differentiate you from the rest. Having said all that, if you don't play the same game everyone else is playing, then you fall behind too focused on these academic assignments when everyone else is focusing on side projects, networking, and working on their career instead of assignments.
So this quote is actually true for students too.
And actually, when you get to the real world, the quote is inaccurate. It should actually be
Companies don't want juniors using AI, in fact it's a red flag cuz they won't learn proper fundamentals to grow. In fact, do most companies even want juniors anymore? Tough times for CS students.
You asked one question right "do most companies even want juniors anymore?" That's why colleges are introducing AI in curriculums. Will they replace seniors, may be yes in some cases.
Companies also don't want to use AI beyond a level where their proprietary and intellectual property is at risk.
AI fails to provide privacy and security as it can't preform great when deployed standalone, only cloud models perform better and companies don't trust them. AI also failed at critical thinking and wide adoption even with local models in place. AI can build apps but not well performing and secure apps, AI works best only with humans in the loop.
AI is also being used by hackers to create sophisticated cyber attacks.
There are always two sides to the coin. AI spending is the main reason for layoffs. Not less need for tech people.
You’ve worded everything right, I do think eventually, when AI isn’t reliable anymore, it would differentiate between people who actually “know” what they are doing and those who relied solely on AI. I do see AI to be an advanced search engine that helps you, but to rely solely on it will be a dangerous ball game all together.
Honestly, I started way before the whole AI/vibe coding thing. My college syllabus was stuck in the past, so I basically had to scrap everything I learned and start over on my own. It was a grind to bridge that gap to modern dev work, but that struggle is actually what got me hooked. There’s something about finally getting real engineering right that just clicked for me.
I learned programming with Pascal, and while it definitely wasn't all applicable to modern development work, the parallels are still there.
IT is just too fast-moving to even attempt to teach people everything they are going to need on the job, so it's more about teaching the mindset that can then be applied to whatever the field is currently up to.
I feel that 100%. The gap between the college syllabus and actual production work is massive. It feels like you have to learn two curriculums: one for the grades and one for the actual job. Respect for the grind that 'click' moment is definitely what keeps us all going.
I think thats a reasonable take, it's a bit of both, we're still figuring out where Vibe coding sits as a technique.
I studied C++ in my degree, but I have always worked in higher level languages - Java / C# / Node etc. I am glad I have had the experience on a lower level language, it helps me appreciate when higher level languages reach their limit how to fiddle with their internals to get what i need to run.
I see vibe coding in a similar manner, over time I'd expect the industry switches to a mode where a majority of engineers do not spend a lot of time writing code by hand, but they know how to do it for the 10% of times when a fine tuned dedicated amount of concentration is required.
This is a great take. It’s not about replacing coding, it’s about shifting where we spend our energy. Having that 'low-level' understanding is definitely the safety net for when the 'vibe' doesn't match the requirements.
chatgpt was released when I was in my second year of my cs degree and instantly I could feel the shift that happened in my brain since then, I no longer would have to force myself to understand smth in order to be able to get the assignment done before its due date, I could just talk with chatgpt and have it give me the code in literally no time and I'd just walk-through it to make sure I understood everything before the discussion and it made me feel like I was so smart while knowing that if I were to try to write the same code on my own I wouldn't be able to.
rn I'm trying to undo the past 2 years of
almost complete dependence on ai but it's not easy ofc but at least I'm trying so hopefully it'll pay off even if it takes some more time
Yeah the downside of depending totally on ai is the fact that you won’t be able to do anything on your own… which is quite scary. But I’m sure you’ll do great getting rid of that dependency. Rooting for you!
totally agree!! and thank you sm:)
Unpopular opinion nobody asked here...
For beginners my take is radical: don't use AI to generate code.
Use it as your teaching assistant (to quiz you, generate test cases, explain a tricky code block). Otherwise you'd be like sending someone else to the gym and then complaining when you don't see your muscles growing.
Never said somebody asked… I think people need to understand that blogs are there for free speech. You don’t need to be rude and say nobody asked, if you don’t like the post, you can just ignore it and scroll I guess
Maame, I enjoyed your post. You raised a really good point. I started my comment with "unpopular opinion" because unlike most opinions I think we shouldn't use AI until we're comfortable coding on our own. Sorry if my comment landed on a different tone. Again, great post.
I guess i misunderstood then, thank you
It’s beautiful code. And I have absolutely no idea how it works.This is the biggest problem ppl have. Using AI to scaffold the code is perfectly fine. It speeds things up. But then you need to reverse-engineer the result and take the time to understand it. If you have questions or doubts, let AI explain. Demand sources for the suspicious claims. This is how I am trying to move forward.
You have perfectly put into words what I feel everyday as a final year student. I do this exact thing which you mentioned that tell LLMs to explain it to me. But still there is another layer of imposter syndrome which i feel. Even though i end up understanding what the task was, I ask myself, "Would I be able to recreate it from scratch?" And the answer is often no, and leaves me a bit defeated.....
I’m starting to realize that understanding the logic is the first hurdle, and syntax retrieval is a completely different muscle we’re still building. You aren't alone in this at all! We're basically learning to "read" the language fluently before we can "write" it, i guess 😊.
Very true! As it is always said... baby steps...
Where there times when you reverse vibe checked that you found out the AI was wrong explaining things?
Yes, constantly. That’s the whole problem with vibe coding—you can't just trust the vibe; you have to verify the syntax. If you don't understand the code, you won't know it's broken until it crashes.
You are not very far from industry, most of engineers I have met recently, they are as good as their copilot is. If you start having arguments they say, “we have delivered what they have asked for in a set timeline “. When industry leaders mentioned their 30% code is written by AI, this means all the redundant code which was written by an engineer has been shifted to AI, but if you are solving real problems , that is still own for computer science engineers/scientists. AI can be assistant not novel thinker that’s my pinch of salt.
You’ve said it all
Some comments may only be visible to logged-in visitors. Sign in to view all comments.