I'm curious about how often everyone is using ChatGPT or other Large Language Models (LLMs) like it in their work. Personally, I've started using it for brainstorming and enhancing my coding abilities, but I've noticed I'm sometimes just copying the code it suggests. How does your team integrate LLMs in your workflow? What tips do you have to effectively use them without losing the learning process?
8 Answers
LLMs like ChatGPT are fantastic for code brainstorming and debugging! I use them daily, but I always make sure to review and test the code before using it. It's best to think of them as starting points rather than just copy-pasting. One thing that really helped me was modifying ChatGPT's output line by line to grasp the underlying concepts—it's a good way to learn!
This advice is exactly what I needed. I've been worried about using LLMs while building my app from scratch. I'm trying not to feel bad about relying on it!
Not using ChatGPT at all. Tried it once just for fun, but it didn't suit my style.
I mainly use it to generate configuration files and handle regex tasks. That's pretty much it for me.
I use ChatGPT mainly for debugging and finding solutions, though I’ve had mixed results recently. But for regular coding tasks, I still put in the hard work myself!
I lead a tech team and we pretty much all use tools like Copilot daily for problem-solving, code debugging, and writing tests. However, there's a lot of concern from leadership about using AI too much to cut staff, which creates a weird pressure. It'll be interesting to see where this goes!
I rely on AI for writing up boilerplate code and for managing my emails. It feels more like having a personal assistant than a coding partner.
I often reach out to LLMs when tackling new coding concepts. For instance, I recently refactored a basic chat server's message cache into a circular one with GPT's help. It's not just about copying suggestions; it helps me clarify implementation and integration, which is super useful!
Yep, I use ChatGPT almost every day! It really helps with brainstorming, debugging, and speeding up boring, repetitive tasks.
I wonder how far you let LLMs take over tasks like documentation. Seems risky without enough human oversight.
I totally agree! I can relate to the struggle of understanding changes without fully grasping the implications. My codebase is a mix too, and I've fixed many messy parts thanks to using LLMs.