Everyone's either saying AI will replace developers or that it's useless hype. Here's what it actually looks like to use AI tools daily as a student developer building real projects.
Suprim Khatri
Backend Developer · February 28, 2025
Scroll tech Twitter for five minutes and you'll find two camps: people saying AI is going to replace software engineers within the year, and people saying AI-generated code is garbage and real developers don't need it.
Both are wrong. Neither is useful.
Here's what actually using AI tools as a developer looks like at least for me.
I use Claude mostly for working through concepts and code I'm stuck on. Not to generate entire features to understand why something works the way it does, to get unstuck when I've been staring at an error for too long, or to get a second opinion on an approach.
For tab completions in my editor, I use Antigravity. It's fast, it's context-aware, and it saves keystrokes on the boilerplate I'd be writing anyway.
For frontend UI, I'll be honest: I use AI heavily. I don't enjoy sitting and writing CSS for three hours. If I can describe a component and get a solid starting point, I will. The key word is starting point I always go in after and clean it up, adjust it, make it actually fit what I need.
For backend, I almost never let AI write code without understanding it first. I write it myself, then ask for feedback on better patterns. That order matters.
There's a version of using AI tools that makes you worse at your job over time. Copy paste a solution, ship it, never understand why it works, repeat. The next time you hit a similar problem, you're back to square one.
Then there's the version where AI is a tool in the same way Stack Overflow is a tool. You use it to get unstuck, to see how something is done, to validate your thinking. But you're always the one driving. You understand what you're building.
The test is simple: if the AI output disappeared tomorrow, could you still build what you're building? If the answer is no, you're relying on it. If the answer is yes, you're using it.
The thing that actually worries me isn't AI replacing developers. It's AI making it possible to seem like a developer without being one.
If you can generate a working CRUD API without understanding HTTP, databases, or error handling you can ship. You can get hired, maybe. You can do work that looks real. Until something breaks in production at 2am and nobody on the team actually understands the codebase.
This is why I force myself to understand what I'm building. Not because AI is bad, but because the understanding is the actual skill. The code is just the output.
AI tools are making good developers faster. They're making it easier for people to start. They're lowering the barrier to shipping something.
They're not replacing the need to understand systems, debug problems, make architectural decisions, or write code that other humans can maintain.
The developers who will be fine are the ones building real understanding alongside the tools. The ones who will struggle are the ones using tools as a substitute for understanding.
I'm a second-year BCA student doing small client gigs, learning Go, building projects I want on my portfolio. I use AI tools every day. They make me faster.
But when something breaks and things always break I'm the one who has to understand why. No AI fixes that for you. And honestly, that moment of figuring it out yourself is still the best part of this whole thing.