4d ago

Efficiency

Programmer Humor @lemmy.ml

Efficiency

67 15

14 comments

3 hours of debugging can save you 30 minutes of reading the documentation
I suppose people will always do what they think is easiest

Vibe coders be like

Eh, I rather write code by hand no matter how long it takes.

Spent a full week trying to get Claude/Cursor to unfuck my router after I asked it to help me set up some isolated subnets via SSH. It really struggled with retaining “object permanence”, or remembering instructions. I had to nuke it and start from scratch…twice.
- Yeah, get too far in or give it too much to start with, it can't handle it. You can see this with visual generators. "Where's the lollypop in its hand? Try again... Okay now you forgot about the top hat."
  Have to treat them like simple interns that will do anything to please rather than admit the task is too complex or they've forgotten what they were meant to do.
  
  Who says you can only have just one on one project? Itemize your tasks in a focused list and divvy up the task to delegates and you can keep the tasks simple enough and coordinate multiple agents on a single Advanced project that one conscience or entity wouldn't be able to do alone. Fluffy ways of referring to tokens AI llm all that

It’s fine if you need to slam out some dirty HTML and CSS. Not so much if you have real problems to solve.

I use Claude for SQL and PowerQuery whenever I brain fart.
There's more usefulness in reading its explanation than its code, though. It's like bouncing ideas back off someone except you're the one that can actually code them. Never bother copying it's code unless it's a really basic request that's quicker to type than to code.
Bad quality and mass quantity in is obviously much quicker for LLMs and people that don't understand the tech behind AI don't understand this actually what's going on, so it's "magic". A GPT is fundamentally quite simple and produces simple results full of potential issues, combine that with poor training quality and "gross". There's minimal check iterations it can do and how would it even do them when it's knowledge base is more bullshit than it is quality?
Truth is it will be years before AI can reliably code. Training for that requires building a large knowledge base of refined working solutions covering many scenarios, with explanation, to train off. It'd take longer for AI to self-learn these too without significant input from the trainer.
Right now you can prompt the same thing six times and hope it manages a valid solution in one. Or just code it yourself.
- Same with writing and image generation. It can give you ideas or handle little details like making sure all your commas are in the right place, the formatting is cohesive, and that you used the right your / you're, or filling in grass or sky textures in the background or putting a bit of polish on a finished image but it definitely requires some editing to get a truly cohesive final result.
- Tbh it would be easier if we could train our own small models on controlled codebases and documentation, instead of random stuff that some people do
  
  We can, but it's a lot of effort and time. Good AI requires a lot of patience and specificity.
  I've sort of accepted the gimmick of LLMs being a bit of a plateau in training. It has always been that we teach AI to learn, but currently the public has been exposed to what they perceive to be magic and that's "good enough". Like, being wrong so often due to bad information, bad interpretation of information, and bias within information is acceptable now, apparently. So teaching to learn isn't a high mainstream priority compared to throwing in mass information instead—it's far less exciting working on infrastructure.
  But here's the cool thing about AI, it's pretty fucking easy to learn. If you have patience and creativity to put toward training, you can do what you want. Give it a crack! But always be working on refining it. I'm sure out there right now someone's been inspired enough to do what you're talking about and in a few years of tears and insane electricity bills, there'll be a viable model.

CW be like:
Spend 7 weeks learning C# (we learned Java in Uni, it's not that hard ffs) and implementing a proper base for a project: Nah
Let ChatGPT generate fucky code and let $me fix it over the span of 7 weeks: Hell yeah
Had I realized that it all was generated, and he didn't have a single little clue how it works, I would've just rewritten it with django or something. Hell, technically the whole server part wasn't needed, it could've been index.html, style.css and scripts.js and that's it.

A nice thing about vibe coding is it can do it in the background while I do other things. I’m not staring at it generating code for 3 hours.
- Agreed. Was stuck in a meeting and let it generate a solution to work on after the meeting.

14 comments