this post was submitted on 18 Jul 2023
66 points (98.5% liked)

Technology

59675 readers
3217 users here now

This is a most excellent place for technology news and articles.


Our Rules


  1. Follow the lemmy.world rules.
  2. Only tech related content.
  3. Be excellent to each another!
  4. Mod approved content bots can post up to 10 articles per day.
  5. Threads asking for personal tech support may be deleted.
  6. Politics threads may be removed.
  7. No memes allowed as posts, OK to post as comments.
  8. Only approved bots from the list below, to ask if your bot can be added please contact us.
  9. Check for duplicates before posting, duplicates may be removed

Approved Bots


founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
[–] [email protected] 26 points 1 year ago (2 children)

Today, I actually disabled copilot in vs code. Yes its impressive at first and sometimes it does save you some typing.

But you'll have to review everything it suggests, and sometimes it will suggest a lot. It's okay code maybe 20% of the time. The rest of the time it's either just doing regular completion, copying and pasting my own code from a few lines up, or just generating gibberish.

Often it will generate things that look okay on first glance, but a deeper look shows significant issues with what it created. Which is a problem often enough that it means you have to deeply look at everything.

I've even had it generate code with typos in it...

It's really not worth the money, your job just changes from person that types something, to person that reviews the robots crap.

[–] [email protected] 4 points 1 year ago* (last edited 1 year ago)

I’ve noticed the same thing with ChatGPT when I ask it to write some code. On first glance, it seems awesome and like it solved your question with ease. However, when you try to go forward and use it, or just look a bit more closely, you realize it has used classes or functions that don’t exist in the library it imports and it made up something. Or, it’s just wrong and not solving the question. This seems fine, but if you go three or four rounds trying to correct it with prompts, and it still gets it wrong, you start to look for other answers.

It’s often helpful for getting headed in the right direction and it has saved me some time, but it doesn’t write flawless code. It needs to be reviewed and corrected by a human.

[–] [email protected] 3 points 1 year ago

Dude same. Normal autocomplete is short and has a low mental overhead but high payoff in time saving from tying. I lasted about 20 min testing copilot before canceling my trial. It slowed me down so much because as you said it generates such large snippets and you have to scan each one to see if it’s useful. Also the constantly flickering of big code blocks changing was super distracting.

I’ve used GPT4 directly for code stuff before and it’s been useful for certain use cases but I find Copilot to be worse than useless in that it not only didn’t really help me it slowed me down and distracted me so much it was a detriment to my coding process.

Maybe Copilot X will change that since it’s basically embedded GPT4 I think, but regular Copilot? Totally worthless IMO