#is it bad if i use AI for coding?

12 messages · Page 1 of 1 (latest)

sterile night
#

currently, how i code/try to code is by writing code(mostly js text turn based games and whatever they require or targeted drills like nested loops, classes with multiple methods, recursions), tryna debug, i suck at debugging, so most of the times i js ask chatgpt what's the fix, like yea i try to understand it to my best of capabilities and writing it down in my own words in my notebook, but i often find myself coming back to chatgpt for fixing my shortcomings

dark oracle
#

kinda answered your own question already, no?

ebon python
#

Yes and no...
Should you use gipity to help you understand and learn why your code wasn't working? Of course. Gipity, much like any coding forum, is a tool meant to help you learn and grow.

Should you depend on gipity so much so that you can't do anything on your own without its input? NO. This not only creates a very strong dependency on it, but there's been plenty of studies already done that prove that you lose the ability to think for yourself

steady path
#

I love vibe coding for local low security tasks that would take me longer to find and deploy some fancy solution.
A good example was logging blood pressure readings.
I could just use a spreadsheet?
I could give the spreadsheet rules to prevent bad entry?
But I don't know how to tell a spreadsheet to move to the next field when the current field has a valid entry and how to ignore me hitting enter when it has auto advanced a field?
I also don't know how to tell the spreadsheet to move forward at midnight to the next date entry?
So I gave my parameters to an LLM and poof I've got a local Electron app that looks great and stores the data in a local SQLite db that's easy to view/graph/share. 🪄

sterile night
ebon python
# sterile night could you give me an example of how either one would look like in practice 
Should you use gipity to help you understand and learn why your code wasn't working? Of course. Gipity, much like any coding forum, is a tool meant to help you learn and grow.

An example of this would be if you are debugging and you know where the issue is, but you don't understand why it's an issue, or you don't understand why a certain response may fix the bug vs another response. Think a bit of stack overflow, or any other coding forum. Majority of the time when you ask a question, you don't just get an answer, you maight also get an explanation of why and how it fixes your issue. This not only helps you solve the current bug, but it gives you a sort of insight which helps you in the long run

ebon python
# sterile night could you give me an example of how either one would look like in practice 
Should you depend on gipity so much so that you can't do anything on your own without its input? NO. This not only creates a very strong dependency on it, but there's been plenty of studies already done that prove that you lose the ability to think for yourself

Although a good tool, you shouldn't just rely on gipity for coding all the time. Isntead of just asking for a quick answer, ask to explain why it works. Asking over and over for just an answer, will only just give you an answer. You will never learn anything, thus, you will never move higher in your career. Use it too much for answers, and you begin to lose the ability to search it on your own

wanton cosmos
#

Dude just use AI you’ll still learn if you put in effort

#

Like if you see the same bug over and over you’ll remember it

#

If you are stuck on the same bug you will remember it

steady path
#

I just vibe coded a tool for my team because the team lead was suggesting a spammy WP plugin to solve a very simple task, but before I could gloat I was instantly faced with having to fix a few minor annoyances with the way it works, started to hate myself for vibe coding it ... and then I just said to heck with it, went back to the LLM and made it fix its own crap. 🛷

grim oriole
#

This is all so new to most people (though a small number of people globally have been using it for a long time). I think standards and best practices will evolve at every level from professional through educational and casual coding. Being with other people at your level and having teacher/TA, team lead, lead dev, etc. will all be great guidelines, but yea it requires a mix still. If the machines can do all the work, that's when we become useless. One neat thing to start thinking about is which tasks are done by AI. For example, a lot of my testing and documentation is done by ai now. However, careful. Functional testing is a core way of developing, such as pytest write the test first. Kind of an old method, will evolve.