Hey everyone! 👋
I'm Nathishwar, an AI/ML enthusiast and a final-year engineering student. Like many of you, I juggle coursework, internships, and personal projects – all while trying to keep up with the fast-evolving tech world. One of my biggest challenges? Coding assignments.
Don't get me wrong – I love coding. But let's be honest, when you're stuck with 3 deadlines, a machine learning project, and a brain running on 3 hours of sleep and too much caffeine... you need help. That's when I decided to try something new.
Here's my real, unfiltered experience with using ChatGPT for my coding work! 👇
😎 Enter: ChatGPT
I'd heard a lot about ChatGPT from my friends and LinkedIn posts. Some used it for writing essays, others for debugging Python code. So, I thought, why not give it a shot?
🔹 My first prompt:
I opened ChatGPT and typed something simple: "Write a Python script to detect prime numbers between 1 and 100."
Boom. Instant result. Neat, readable code with a short explanation.
🧠 Initial impression:
Not gonna lie – I was impressed. But that was just the beginning.
🧠 The Real Test – My ML Mini Project
For one of my AI courses, I had to build a handwritten digit recognition model using CNN. Normally, this would take me hours – setting up the dataset, writing the model, tweaking layers, and handling bugs.
🔹 The challenge:
I asked ChatGPT to: "Create a CNN model using TensorFlow for MNIST digit classification."
And within seconds, it gave me a fully functional script – including data loading, model architecture, training, and evaluation.
🔍 What I actually did:
Now, I didn't just copy-paste and submit (that's not how you learn). But it gave me a solid base to build on. I tweaked the hyperparameters, added dropout layers, and even improved accuracy by 2%. It felt like pair-programming with a superfast AI buddy.
❤️ What I Loved
After a few weeks of using ChatGPT consistently, I noticed several benefits:
🔹 Clear explanations:
I often asked why it used ReLU or what a certain function did – and got short, understandable answers.
🔹 Debugging help:
You know those vague Python errors that make no sense? I pasted one into ChatGPT and got not only the fix but why the error occurred.
🔹 Time efficiency:
Instead of googling 10 Stack Overflow pages, I got focused, tailored answers.
😅 What I Learned the Hard Way
But it wasn't all sunshine and semicolons.
🔹 The limitations:
- • Sometimes, it hallucinated. A few times, it gave me code that looked legit but didn't run. It used non-existent functions or outdated syntax.
- • It can't read your mind. I had to be specific with my prompts. "Make it faster" didn't help, but "Optimize my loop with NumPy" did.
- • Blind copying = bad idea. If you treat it like a crutch, you stop thinking. I learned to use it as a guide, not a replacement for understanding.
💭 Final Thoughts
Using ChatGPT for my coding assignments was honestly a game-changer. It didn't just help me finish tasks faster – it made me a better learner. I spent less time stuck and more time experimenting.
I started asking better questions. I even got curious about how ChatGPT works, which led me deeper into NLP and transformer models.
So, if you're a student like me, here's my advice: Use ChatGPT smartly. Don't let it replace your brain – let it level you up.
Happy coding! 🚀
- Nathishwar
"The best way to learn is not by memorizing solutions, but by understanding processes. AI tools can be your guide, not your replacement."
— Anonymous Computer Science Professor
