A key part of the definition of vibe coding is that the user accepts code without full understanding.[1] AI researcher Simon Willison said: "If an LLM wrote every line of your code, but you've reviewed, tested, and understood it all, that's not vibe coding in my book—that's using an LLM as a typing assistant."[1]

Did we make it from AI hype to AI dunk in the space of a single Wikipedia article? Lol.

[–] Ephera@lemmy.ml 42 points 2 days ago (1 children)

Interesting that the term was coined by someone who presumably intended it to mean a good thing. I assumed it to be an entirely derogatory term...

[–] Ropianos@feddit.org 15 points 1 day ago* (last edited 1 day ago) (3 children)

Well, lots of words are sort of derogatory in programming right? Hackers hacking things together, rubber duck debugging (you have a bug? Go talk to a toy!), git blame and probably more

Edit: forget git blame, git itself is already slang for idiot. As Linus Torvalds once said, he created two successful projects, both named after himself

[–] Ephera@lemmy.ml 6 points 1 day ago (1 children)

Sure, but those words have been coined by the nerds, or those that don't try to sell you anything at least. The guy very much wants to sell you vibe coding and LLMs and whatnot.

[–] Ropianos@feddit.org 5 points 1 day ago

Andrej Karpathy is absolutely a nerd to me. He's not about selling you LLMs but about teaching you how to build your own (I recommend checking his videos out, they are one of the best resources on LLMs)

[–] ulterno@programming.dev 3 points 1 day ago

And reflog!

Oh! It was reflog, sry.

[–] AnotherPenguin@programming.dev 3 points 1 day ago

Anything can be derogatory if you try hard enough

[–] sturlabragason@lemmy.world 9 points 2 days ago

Dude, love ❤️

[–] Mirror Giraffe@piefed.social 6 points 1 day ago (1 children)

I've ended up becoming the sole caretaker of two react native apps, something I did not ask for but some people were fried.

I'm not going to manually troubleshoot and learn everything that breaks when I need to update a dependency. I'll vibe through those errors, learn what I have time for and then test it like mad.

A lot of the time I end up having to solve things myself but my job wouldn't be tolerable if I had to manually work that shit.

[–] MajorHavoc@programming.dev 3 points 15 hours ago* (last edited 15 hours ago)

Yeah. The "this got dumped on us and we're doing the minimum until we can replace it" is a genuinely solid use case for vibe coding.

And honestly, that's all I usually did with those before AI came along anyway. So I welcome better tools for it.

[–] jmcs@discuss.tchncs.de 91 points 2 days ago* (last edited 2 days ago) (1 children)

Just wait until the vibe coder overwrites 3 months of "work" with garbage because ChatGPT never told them about git and then decided to poop the pants.

[–] Binette@lemmy.ml 13 points 1 day ago

already happened lol https://lemmy.eco.br/post/11724141

[–] jet@hackertalks.com 65 points 2 days ago (3 children)

Now... Now... Wait until they have to debug it

[–] heavydust@sh.itjust.works 52 points 2 days ago (2 children)

Or fix a simple syntax error. I've seen that a few times from juniors who thought they were too smart to learn, and it's painful to watch.

[–] scrubbles@poptalk.scrubbles.tech 10 points 2 days ago

Yup, seeing that too. They're all proud of the work they didn't do, but the second there's a syntax error they panic.

[–] jackalope@lemmy.ml 5 points 2 days ago (1 children)

LLMs excel at fixing typos. That's honestly the most useful aspect of them.

[–] Excrubulent@slrpnk.net 1 points 16 hours ago

Which is also not a real use-case. Like, just check your work, you don't need a tool for that.

Hell I don't even use spell-check because it's just distracting, and it's a fundamentally wrong way to think about language.

[–] magic_lobster_party@fedia.io 31 points 2 days ago (1 children)

They will resort to vibe debugging

[–] Artyom@lemm.ee 6 points 1 day ago

"I'm a full stack Viber. Vibes all the way down"

[–] iAvicenna@lemmy.world 5 points 1 day ago

which is like %80 of the time

[–] lowleveldata@lemmy.world 35 points 2 days ago (1 children)

idk man. Algorithms are always the fun part for me. Real work, not so much.

[–] theneverfox@pawb.social 35 points 2 days ago (1 children)

I love the in between, where you have to actually adapt algorithms to useful situations

It's like 2% of all the work I do, but those moments stick with me

[–] BleatingZombie@lemmy.world 5 points 2 days ago (1 children)

I agree. But there's never much time for it

[–] theneverfox@pawb.social 3 points 1 day ago

What do you mean? It's a problem in your way, it takes how long it takes to get through it

[–] xthexder@l.sw0.com 35 points 1 day ago

If those vibe coders knew what a binary tree was, they'd be very upset.

[–] TedDallas@programming.dev 27 points 1 day ago (1 children)

Vibe Coding essentially automates copy-and-pasting the answers from StackOverflow.com.

But sometimes it pastes in the code from the questions.

[–] TechieDamien@lemmy.ml 9 points 1 day ago

To be fair, sometimes it is right to take the code from the questions. Eg if you want behaviour x and the question is "how do I do y, my code is only doing x?" Then the code you want is from the question.

[–] Ephera@lemmy.ml 13 points 2 days ago (2 children)

I mean, surely you did it for the exercise, right? If you needed a solution, you could've very likely used a library...

[–] chicken@lemmy.dbzer0.com 5 points 1 day ago* (last edited 1 day ago) (1 children)

Having fewer/no dependencies is nice though

[–] Ephera@lemmy.ml 2 points 1 day ago (1 children)

Well, we have a tool for that called vendoring, a.k.a. copy-pasting the library code into your repo. It's no worse than copy-pasting LLM-generated code...

[–] chicken@lemmy.dbzer0.com 2 points 1 day ago* (last edited 1 day ago)

To me the disadvantage would be, the library likely does many more things than just what you need it for, so there is way more code, so you probably can't realistically read and understand it yourself before incorporating it. This would lead to among other issues the main thing that irritates me about libraries; if it turns out something in it is broken, you are stuck with a much bigger debugging problem where you first have to figure out how someone else's code is structured.

Although I guess that doesn't apply as much to implementations of common algorithms like OP since the library is probably solid. I would consider favoring LLM code over most anything off npm though.

[–] ulterno@programming.dev 4 points 1 day ago

That screaming guy made the library.

[–] hddsx@lemmy.ca 9 points 2 days ago (1 children)

Vibe coder?

[–] rikudou@lemmings.world 18 points 1 day ago* (last edited 1 day ago) (1 children)

Stupid term people use for "writing" code without the knowledge of programming by using AI.

Kinda the next level of running random code from the internet without understanding what it does.

[–] hddsx@lemmy.ca 12 points 1 day ago (3 children)

Maybe I’m old, but I avoid AI specifically because I’ve seen the quality of code out there (not to say my quality is amazing), and if AI learned from that…. Well

[–] yucandu@lemmy.world 10 points 1 day ago (2 children)

AI is great for when you're typing out 30 slight variations of the same thing, and you can just be like "see what I'm doing here? Do it for the other 30 variables" and it does it just fine.

[–] rikudou@lemmings.world 6 points 1 day ago

It's also great if you have a general knowledge of something but don't know the details. Like today I needed to do some database introspection using queries in Snowflake, I knew exactly what I needed but not where the database schema is located etc., so I let GPT write the query instead.

Or some time ago I needed to get all instances of classes implementing a specific generic interface in .NET, the code eventually dabbled into the very specifics of the runtime, it would've taken me much longer to find out with documentation.

All in all, it's my opinion that AI is great if two conditions are met:

you know exactly what you want to do and you can specify it to very tiny details
you have the knowledge to verify whether the result makes sense without running the code (or at least the knowledge that it can't break your app or computer)

[–] zenpocalypse@lemm.ee 2 points 1 day ago

*pastes list of 100 numbers"

"Flunky, put this in a where in SQL statement. "

[–] Aurenkin@sh.itjust.works 3 points 1 day ago

I don't think it's necessary to strictly avoid it by any means, it's quite a useful tool. You just have to use it appropriately, similar to stack overflow in some ways. That said, obviously you should use / avoid whatever tools as you want.

[–] zenpocalypse@lemm.ee 2 points 1 day ago

It's taught me a few new tricks. It's good at short bits of code, but anything longer is likely to have problems that take as much time to fix as writing it yourself to begin with.

I also find learning something new to be much faster when you can just ask a question about best practices or why something works like it does.

[–] someguy3@lemmy.world 8 points 2 days ago (1 children)

Can someone explain?

[–] addie@feddit.uk 20 points 2 days ago (1 children)

A binary tree is one way of preparing data, usually for sorting. Each node can have a left, right, or both, children.

  A
 / \
B   C
   / \
  D   E

"Inverting the tree" means swapping the children for each node, so that the order that the nodes are visited is reversed. Depending on whether you want to copy the tree or swap it in place then the algorithm is different. C++ provides iterators too, so providing a "order reversed" iterator can be done efficiently as well.

You're going to have to visit every node and do at least one swap for every node, and an efficient algorithm won't do much more than that. Bring unable to do it suggests that the student programmer doesn't understand stacks or recursion yet, so they've more to learn.

[–] Thorry84@feddit.nl 15 points 2 days ago (1 children)

Important context:

This is often an exercise for beginning programmers, it's a very simple task that's easy to understand, but leaves enough room in the implementation to make it a good exercise.

Sometimes it's used as a test on job applications, which is total bullshit, it isn't a good test of someones actual skills as a software developer. Because of this it's become a bit of a joke on the internet.

[–] barsoap@lemm.ee 4 points 1 day ago* (last edited 1 day ago)

This is often an exercise for beginning programmers

And non-beginners have to weigh the skill of the interviewer to figure out whether "To invert, I'll just keep the tree as it is" is an approach they can grasp. Enlightenment is realising that the tree is already in the perfect shape.

[–] vane@lemmy.world 7 points 1 day ago

...into cursor or windsurf, how the fuck they come up with those names dude ? Vibe coding on windsurf.

[–] wise_pancake@lemmy.ca 7 points 2 days ago (1 children)

I’ve always stayed away from tree algorithms

[–] QuazarOmega@lemy.lol 3 points 1 day ago

I always stay away from trees

[–] xtools@programming.dev 1 points 4 hours ago

vibe coding is the dumbest term they could have come up with. can we call it imbecile coding instead?