sooo i went through this last year when my budget was basically zero lol. honestly i had some issues with the paid stuff cuz i just couldnt justify the monthly sub for sophomore projects... unfortunately GitHub Copilot was kinda mid for learning algorithms imo since it just spits out code. i actually tried using the free version of Tabnine but it wasnt as good as expected for complex java stuff. lately ive been sticking with DeepSeek Coder V2 and it's been highkey great for my python labs cuz it explains the logic better than some paid tools ive tried!

TL;DR from this thread: most people agree that while GitHub Copilot is okay for speed, it's pretty mid for actually learning logic. Everyone seems to prefer using models like Claude 3.5 Sonnet to explain the 'why' behind the code.

I'd be careful though... just copying AI code can lead to security vulnerabilities in your projects. If you want a solid, safe alternative, check out Codeium Individual—it's free for individuals and actually respects your privacy better than most. Definitely worth a look if you're on a student budget but wanna stay safe! peace.

Been thinking about this all afternoon and tbh I was super skeptical about using AI tools at first because I didnt want to lose my own edge as a dev. My big worry was that if I relied on them too much, I wouldnt actually learn how to maintain my own code in the long run. After trying a bunch of stuff, I ended up really liking Amazon Q Developer. Its free for individual use and it feels a lot more secure and reliable than some of the newer stuff popping up. It actually flags potential security issues which is a huge plus when you are still learning the ropes. Another one I keep in my toolkit is the Supermaven Free Tier. The context window is insane, so it can see your whole project at once. Instead of just asking for solutions, I use it to explain how my different classes interact. It really helps with that long-term understanding of your own architecture. My best advice is to use these tools to write unit tests first. If you can use the AI to help you build a test suite, youll actually understand what your code is supposed to do instead of just hoping it works.

Seconded!

Coming back to this, sophomore year is definitely the point where things get real for a CS student. It totally reminds me of my old roommate, Mike. He was so obsessed with the DIY approach to coding tools that he refused to use any IDE or assistant at all. He spent an entire semester trying to build his own custom text editor from scratch using only assembly language because he thought it would make him a purer programmer or something. I would walk into our dorm and see him surrounded by handwritten charts of memory addresses, looking like he hadnt slept in three days. He was so busy debugging his own editors cursor logic that he completely forgot to actually write the sorting algorithms for our Java homework. He ended up failing that algorithms course because his editor kept crashing every time he tried to type a semicolon, but man, he was committed to the bit. It was a wild thing to watch, totally unproductive but pretty legendary in a weird way. Just one of those college stories you never forget...

Did this last week, worked perfectly

Saved for later, ty!

Ok so, I totally feel u on the sophomore year struggle. I remember literally staring at my screen for hours trying to figure out Java data structures last semester, and it was so overwhelming. Tbh, I'm still kinda new to using AI for coding, but I've had a really good experience lately just sticking with the big names. Honestly, for a student budget, you really can't go wrong with anything from Google or the stuff Microsoft puts out.

I mean, I've been using their tools for my Python scripts and they actually help me learn instead of just giving me the answer. Here's what's been working for me:

* Just use any of the tools from Google—they're lowkey amazing for explaining why a bug is happening rather than just fixing it.
* The Microsoft stuff is seriously solid for Java because it integrates so well with the IDEs we use in class.
* Basically, stick to the free tiers of the major brands and you'll be set for most algorithms.

I was sooo happy when I finally got my linked list project working without it feeling like I was cheating. It’s kinda about how you prompt it, you know? Like, I always ask it to explain the logic first. Anyway, I’m super satisfied with just using the standard brand-name extensions for now. They work well and I haven't had any complaints yet! Good luck with your projects, you got this!! 👍

I went through this last year. I spent way too much time relying on auto-complete, and honestly, it lowkey hindered my understanding of algorithms. I switched to using Claude 3.5 Sonnet for explaining logic rather than just writing code. It's basically been a lifesaver for Python debugging.

* Focus on the "why" by asking the AI to explain the time complexity.
* Use tools that provide step-by-step logic for Java structures.

It's easy to get lazy, but forcing yourself to understand the logic first is key. gl!