[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/g/ - Technology


Thread archived.
You cannot reply anymore.


[Advertise on 4chan]


File: sadcowboy.jpg (7 KB, 220x229)
7 KB JPG
>ask vibetards to share what they have made with their supposed expert programmer machines
>they never EVER share anything and just call you a tranny and post weird cat pictures
Damn you're right I feel like my job will be replaced by you any day now!
>>
File: IMG_1466.gif (2.92 MB, 300x240)
2.92 MB GIF
>>109504918
I made this with AI. Took me 798 tokens. 11 days.
>>
>>109504918
How many times will you make this thread this week?
>>
Vibecoders want to be programmers, and insist we call them programmers. But they will never be real programmers. They have no ability, they have no experience, they have no brain. They are AI shitters twisted by prompts and workflows into a crude mockery of nature's perfection.
>>
>>109505821
Promising. Worthy of a full pasta
>>
so OP, will you show us what you've programmed?
>>
>>109504918
>Hi my skin is the color of shit. I demand that you tell me how to use ai to make money saar
>>
File: nooo.jpg (41 KB, 640x596)
41 KB JPG
>>109504918
>NOOO I AM THE ARCHITECT. I AM A REAL PROGRAMMER I KNOW HOW TO CODE. BUT ITS WASTE OF TIME. CLAUDE DOES IT FOR ME I CHECK IT THROUGHLY NOOOOOOOOOOOOOOOOOOO
>>
File: 1698133175402.jpg (335 KB, 1024x1024)
335 KB JPG
>>109504918
why would i share my work with you, tranny? if you wanna be obsolete that's just less competition for me.
>>
Just realized the AI issue can be used to drive a wedge between the corpo-tranny hegemony.
>trannies hate AI because they feel threatened by it
>corpos use it to cut back
Hating AI is the correct position, but terminally online trannies who use FOSS as a surrogate for power and meaning will become worthless.
>>
>>109505971
>why would i share my work with you, tranny?
Yes. Don't share it. World doesn't need any more of your slop.
>>
File: 1769753628602099.jpg (89 KB, 401x358)
89 KB JPG
Not mine but here is the most vibecoded slop I've ever seen
https://github.com/diegosouzapw/OmniRoute
The UI is so shit I couldn't figure out how to stop it from banning Gemini every time google shits itself with 503.
>>
>>109506018
Vibecoded project final boss
>>
File: 1598038695811.gif (105 KB, 157x255)
105 KB GIF
>they never EVER share anything
Today's your lucky day!
FFmpeg (a very impressive project) is vibe-coded: https://x.com/FFmpeg/status/2084084810813743614
Linux Kernel (one of the most important projects out there) is also vibe-coded: https://www.tomshardware.com/software/linux/linux-developers-are-using-ai-vibe-coding-to-keep-vintage-amd-gpus-alive-r600-driver-cleaned-up-with-github-copilot-gives-hd-2000-to-hd-6000-series-a-new-lease-of-life

If you need more examples just ask!
>>
>>109506322
ffmpeg is not vibecoded, it is becoming vibecoded. Same with linux.
Saying ffmpeg is vibecoded is like saying windows was made by an indian man.

Eitherway, as soon as the vibeslopification started, everyone could tell.
>>
>>109506322
Retarded faggot.
>>
>>109506322
Good bait, made me reply
>>
>109506322
>still doesn't share anything
incredible
>>
>>109506322
AI assisted is not the same as vibe coded. Neither of those projects are vibe coded. Literally every single serious project is AI assisted nowadays because it's much more convenient and faster than spending hours googling stuff or writing boilerplate by hand. Still not vibe coded slop
>>
File: 1757177908794059.png (19 KB, 666x107)
19 KB PNG
>>109506322
I don't understand these people, anthropic isn't public so what is this guy's motivation for being like this? it would at least make some sort of sense if he owned shares
>>
>>109506597
He sucks Dario and Tibo for usage resets so he can code his sloppy react todo list app
>>
>>109506322
But what have (You) coded? Don't take credit for real vibeGODS' work, vibeSTOLENVALOR.
>>
>>109504918
True and they don't even post cat pictures they made themselves.
Just the same boring cat pictures they saved months ago.
>>
>>109505027
How did it fuck up the frame order?
>>
>>109504918
I created a simple software that will OCR text from pdfs, or just read them if they have text and output a clean ePub file for reading on my super old kindle.

I didn't have it take pictures because I was planning on just taking books.

Works fine keeps formatting, and it has a optional spell check passage to correct issue.
>>
>>109506756
I think it's really cool that LLMs made programming accessible and that anyone can easily and quickly make things like these. That is really not that complex tho. The hard parts, such as the spell check, the OCR, or the GUI library have been imported from existing packages. The LLM just put some (presumably python) glue code between those components
>>
>>109506585
t. nodev
AI slop is completely worthless in a professional setting.
>>
>>109506869
are you seriously implying that just asking an llm in a VS code chat window a question about a popular library is not on average much faster than looking at docs?
>>
>>109506893
Yes, because it always messes it up, including but not limited to:
- Making up functions that don't exist (yes, to this day)
- Confusing multiple versions
- Confusing similarly named libraries
- Making up details about e.g. parameters of a function (for example, claiming the input is either a filepath or a buffer, when actually it is only a buffer)
- Removing the exact details you want in favor of bullshit """summarization"""
- Refusing to follow your instructions and giving you suggestions about how to find the docs instead of giving you the docs
- And much, much more

Meanwhile the docs for the exact thing I'm looking at is the first link on google. As a bonus, that workflow even works for unpopular libraries!
>>
>>109506893
yes because pajeet's vibe slop (encouraged by mgmt) that coincidentally passes the tests and was approved by another pajeet will cause 3 days of data loss while it's being urgently rewritten the remaining whites in the company
>>
>>109506918
>Meanwhile the docs for the exact thing I'm looking at is the first link on google.
Not for long, snailcat. Soon first pages of all search engines will be filled with useless ai wikis produced by vibegods.
>>
>>109505027
That's pretty shit you should be ashamed of posting this.
>>
File: bootlicker-openai.jpg (32 KB, 440x334)
32 KB JPG
>>109506953
>t.
>>
>>109506953
Unfortunately true, for example most python search results are ai generated incorrect slop while the official python docs are at the end of page 1 or on page 2 somewhere. But it's been like that for ages already, it was spam like quora or geeksforgeeks or other pajeet site before AI.

But for other languages that isn't the case for now.

Also, to be clear, LLMs are excellent for search and that's where they should be used, instead of shoehorning them everywhere but where they excel. No bullshit """summary""" with made-up fake citations pointing to completely irrelevant random sites that don't contain anything being "cited", just use it to identify hits from the database by providing query/filter/rank tools to the LLM and give me the output + "preview fragment" (potentially with a collapsed exact copy-paste from the source, not rewritten by gay eye).
>>
>>109506893
It's the resident schizo who claims to be working for FAANG and claiming to be developing fundamental economy while posting here 24x7 seething about AI. Just ignore him.
>>
>>109508269
Meds, now.
>>
>>109506893
He also denies he's not the same guy while making the same post every thread and claims how he's some AI engineer deploying the largest and latest models and sheit while he obviously don't know how to use them. Notice how in your reply to you, he's talking about a 'paste code in chatgpt tab' workflow while you were asking about harness engineering.
>>
I don't know why anyone cares or even gives this attention. It takes 1 afternoon to evaluate the tools yourself and come to your own conclusion. What do I care if people try to argue one way or the other, I already know the truth first hand.
>>
>>109504918
Why would I share proprietary content on here? You can see us lapping you every day at work. Go back to linkedin and polishing your resume, loser.
>>
Frankly this reminds me of the gear vs automatic debate few years. The argument just comes down to how making things harder for you is just better.
>>
>>109508338
there wouldnt be a debate if there werent advantages to either
>>
>>109508307
You're referring to at least 3 different people, of which only one is me. Hilarious. But really, take your meds.
>>
>>109506893
Speed is not my concern for that question, accuracy is.
In my experience, more often than not, even the latest LLMs are giving me misinformation about basics like syntax or standard features, on top of being 10+ years behind on their knowledge of the language.

Sometimes this is subtly bad too.
If it's wrong about what a language supports, oh well you get a compiler error or maybe that standard library function it used is just made up.
But It's even worse when it produces something syntactically correct, but outdated and/or bad practice.

It's frustrating to interact with. Like having a junior programmer at your disposal except they never learn or get better. They just forget that feature XYZ has been in the language for 5+ years, even after you tell them and they have access to current docs, etc.
>>
>>109508312
More importantly, why are people spending so much time and energy shilling when, as you say, it's so easy for anyone to just try it yourself? It's not like this is different companies' employees being paid to post here or whatever so what the fuck?
>>
>>109508371
>junior
Apologize to juniors worldwide.
Even interns aren't that bad on average.
>>
>>109508380
I suspect most of the marketing is automated and not actual people. That makes the most sense to me considering it's an AI business and the thing they're best at is confidently lying in human language, which is the basis of modern marketing.
>>
>>109508371
This has not been my experience. Frankly, from my experience AI is able to pull the latest references and sometimes stuff I didn't know about. Which even the official documentation doesn't contain because it's actually have access to library context.
>>
>>109508307
It's not one person, and while people that completely deny the usefulness of LLMs are dumb, I also agree that most vibe coders suffer severely from the dunning kruger effect and can only produce trash prototypes. While I use LLMs in the IDE, my workflow is similar to the "copy paste from a web chat" one. Letting the LLM write directly to the file incentivizes me to check the changes lazily. I prefer to put the LLM in ask mode and copy in the snippets myself, that way I really do read everything and it's only an extra second
>>
File: IMG_2176.jpg (117 KB, 828x1160)
117 KB JPG
>>109506735
What I cant do, is tell you why. What I can do, is tell you it knows best.

>>109506961
15 sudanese orphans died from dehydration to make that.
>>
>>109508380
>why are people spending so much time and energy shilling

Why do you assume it's shilling and not people genuinely impressed by it and advocating it's usage?
>>
>>109508428
nta but
-its because they almost never produce any example of their work
-its because they attribute behaviour to the llms that cannot be replicated
-theyre closeted homosexuals. this may or may not be a relevant epidemiological factor
>>
>>109508417
>I really do read everything
This I think is critical and how I get the most utility out of the machine.
However, I question the efficacy of it most of the time.
It's not bad to consult with the machine, but I feel like the fact that I have to babysit and correct the majority of the output, is equal or more time than just doing things myself. Most importantly it takes away from the autonomy, and the autonomy feels like the only real way to gain back time.
Right now I can run several agents in parallel and have them do a lot, but I can't realistically trust any of that. Which means I either need to read and correct it in-process or make a review all at once at the end, at which point I'm sinking manual time into the project anyway.

The fact that's its not a clear decision as to whether this is a waste of time or not, is probably a good sign. It's not clearly bad, and can improve, but I feel like most of my sessions I leave thinking "I could have probably done that faster by hand from scratch that refining the slop by hand into something usable".
>>
I think you're a loser if you hate AI. AI is just a tool. It simply is. And it saves time. AI for lack of a better word, is good. AI is right, AI works. AI clarifies, cuts through and captures the essence of the evolutionary spirit. AI in all of its forms. AI for life, money, love, knowledge, has marked the upward surge of mankind, and AI - you mark my words – will not only save you, but that other malfunctioning corporation called the USA
>>
>>109508417
>It's not one person
Their posts are surprisingly repetitive that its fair to assume it's just one person or at most 2 or 3 copycats.
>I also agree that most vibe coders suffer severely from the dunning kruger effect
But the poster wasn't replying to a vibe coder cat poster, but someone who was talking about AI assisted coding. This is just a motte bailey fallacy.
>my workflow is similar to the "copy paste from a web chat" one
And that's where most of the criticism about AI coding comes from? Without proper context awareness, it's bound to produce incorrect output except for the simplest stuff.
>>
>>109506018
Beastly
>>
>>109506597
Gootecks is a professional retard ignore everything he does or says
>>
>>109508468
I use it in the IDE, it has all the context it needs. That's the reason I use it in the IDE
>>
>>109508468
>its fair to assume
It is in fact hasty and not fair at all.
Come to understand that people of same ilk gravitate towards same places of discussion.
It's even worse since some people treat 4chan as a clubhouse (it's not) where there are fashion rules on how you must write (there aren't any actual style rules like that, only site rules).
>>
>>109508451
I've seen you make this thread many times this week, every time someone gives an example, you almost always move the goalpost, either by redefining what counts as 'software' by your standards or redefining what counts as 'made with AI'.
It can be tiring talking to a brick wall. Hence why you're getting lesser and lesser (You)s every time you make this thread.
>>
>>109508485
nope. wrong number.
when im making a thread it gets nuked within 5 minutes because its clear-cut shitposting

also its childish to care about you's beyond them being a means to have a coherent discussion on an anonymous board without ids
>>
>>109508460
>I feel like the fact that I have to babysit and correct the majority of the output, is equal or more time than just doing things myself
This is a very bad criticism of these tools that I can only assume you are either not a dev or someone in upper management ego hardly codes these days. I think people overstate how actually cumbersome it is.
You should be reading everything you wrote, testing, verifying output and all even if you didn't write it using an LLM. This is the gist of most proper software development workflow, not just AI assisted one.
Second of all, this still saves me much more time than if I read the docs myself and wrote everything stroke by stroke.
>>
>>109508548
You're restating what I already said.
I'm saying that I do write and review all my own stuff.
If AI doesn't eliminate that, it doesn't save me any time.
I trust humans a lot more to get a task done with even less of my intervention.

The irony of your remark is that my standards are high because I have been developing software for a long time and the machines are not up to that level yet. Hence why I have to review and refine their work.

The promise being sold is autonomy at the same skill level as a software engineer, and that's not what we actually have at our disposal.
We may have the ability to deploy an AI fleet, but that doesn't mean we have the ability to take advantage of it since we're still limited by a person reviewing all that work.
If I have to rewrite 90% of the function body 90% of the time I may as well do it by hand to begin with.
>>
>>109508460
>>109508548
Also even if you're someone in upper management, you've to do the same step again with juniors and interns(which most don't do, they just assume they did their job and just skims through).
An arrogant dev pushing something to production on Friday night without verification is the reason why most software blunders in history have happened.
>>
>>109508592
>you've to do the same step again with juniors and interns
I addressed this already.
The problem is that in the last several years I have trained multiple people. They learned and got better.
The LLMs simply can not. They know what they know and you must supplement new knowledge for them, and constantly reinforce this.
Junior programmers have the potential to become senior level. LLMs are the level they are, forever.
No amount of time sink pairing with them is going to make them better by the end of the week.
>>
>>109508590
>If AI doesn't eliminate that, it doesn't save me any time
It eliminates the time I've to search, read, search, write boiler plate, test, rewrite boiler plate. The only thing I've to spent my time own is the final implementation of the business logic.
>If I have to rewrite 90% of the function body 90% of the time I may as well do it by hand to begin with.
Again, not my experience, I've to rewrite mostly something like 10-20% of the output, depending on how complex the task is.
>>
>>109508608
>The LLMs simply can not
>LLMs are the level they are, forever.
Again simply wrong. They have improved a lot since they've come to the market.
>>
>>109508628
(nta)
>>claude write me a test
>the test:
int test (void *data)
{
printf(passed.\n);
return (1);
}


tests are the one thing you have to be extra careful about
>>
>>109508628
I wish that was me. The code being produced by any of the models all seems to normalize. Which makes sense they're all learning from the same sources.
The problem is that what they're learning from is mostly mid level garbage.
So even with high standards, guidelines, etc. it just breaks apart so easily so fast even with constant reminders.
>no that fucking ass, that's the worst way to do that and I told you this already

>>109508639
Just because new models come out doesn't mean models can improve. As I said above, no amount of repeating something will make it learn. Even if you go the route of putting instructions on disk and make it read them every step, it still might just sometimes forget or disregard you. You literally can not teach it beyond its training set for longer than its small context window.

That's not the case with people. I can get you up to speed quick and then you stay there.
If I get an LLM up to speed it forgets everything I taught it in 5 minutes and ignores all the notes it took.
>>
>>109508651
>tests are the one thing you have to be extra careful about
>is not careful when writing his example test

point still stands.
>>
>>109508651
You should be writing tests proper before you start development, not after wards.
Most trash tests written by AI is driven by retarded upper management driven test coverage requirements anyways.
>>
>>109508690
>You should be writing tests proper before you start development, not after wards.
it depends on how/what you develop
if youre doing exploratory programming, you wont know what are the contracts you will have to uphold in the end

on the other hand- sure, ideally we should have everything running on paper before even comitting to writing even one single line of code
but sometimes you really dont know what you will need in the end
rn im working on a trading bot which was a text sentiment analyzer a dozen iterations ago...
>>
>>109508690
Writing tests first always felt like crystal ball logic "just implement it right the first time". Yeah that would be great but it's not practical for anything non-trivial.
If you have to change production code during a refactor, it necessitates rewriting the tests anyway. Feels like a waste of time writing 2x the discarded code.
>>
>>109508338
>You literally can not teach it beyond its training set for longer than its small context window.
That is simply not true.
Chatbots today aren't 00s era cleverbots that just gives a canned response. You can either a train a custom model for your requirements or give the necessary context though it is still limited by it's context windows.
>The problem is that what they're learning from is mostly mid level garbage
Most of them are trained on publicly sourced open source code used by many or top voted SO answers. What is your standard for a 'good' source?
>>
>>109508770
Popularity does not equate to quality. Linux is a popular open source project for example.
>>
>>109508770
nta, but viewed through the lens of stats
the best implementations will always be less statistically significant than the canonical ones

and speaking about C, the best implementation is one that synergizes with the data youre working on
and that is information thats just absent from the training data
it learns from repos, not people's computers' contents, and what they apply specific programming patterns to
>>
>>109508730
>>109508746
Then you just rewrite the test cases and update for the new prod. You should've an understanding or at least vague idea of what your output is going to be.
>>
>>109508786
This isn't /a/ or /tv/ where you can just pass such vague value judgements. This is a field where you can compare and verify the results. Most of the time, popularity do coincide with quality.
Here, there's either software that everyone complains about or software that no one uses.
>>
>>109508795
rn, i honestly dont
and its not because of my main project, but because of the underlying lib
i got fed up with building stuff that i *think* will be useful at some point, and ending up with a bloated mess
so now i update and modify my libs as i go

for example: i changed my early exit function 3 times in the past week because its behaviour turned out to be impractical in certain cases
and while i could write down my main project on paper
its... its too fucking big, man. i could but i *really* dont wanna.
>>
>>109508795
The whole point of using these tools is to save time. We want to minimize the amount of time rewriting code not increase it.

>>109508833
Your point works against you here. You're right that we can have objective measurements, and the measurements say that the state of software right now is abysmal. And this is where the machines learn from.

>Most of the time, popularity do coincide with quality
From my perspective it doesn't look that way. It seems like what is popular is what is predatory; has nothing to do with merit. More like vendor lock in.
>>
>>109508786
>>109508833
And most of the time, low quality is usually punished, no matter how gigantic you think the market dominance is, unlike the media industry.
Just look at previous examples of Skype and IE.
>>
>>109508865
You couldn't have picked a worse example.
Skype and IE were literally so popular and successful that Microsoft got an anti-trust lawsuit over their monopoly.
At the same time IE was a laughing stock.
>>
>>109508857
>We want to minimize the amount of time rewriting code not increase it.
You've to rewrite it if the business logic changes as often happens in software development. Also, my point was generally about all test cases writing, not just those written by AI.
>>109508884
You could've not used a worse argument.
And it was still dominating even after the anti trust lawsuit, with it's competitor shutting down.
It was not until a genuinely better product came around.
>>
>>109506322
When people say "vibe coded", they don't mean every project in which there is at least one line of code made with the help of AI. FFmpeg and Linux kernel were "impressive projects" long before vibe coding was even a thing.
>>
>>109508884
>>109508900
*genuinely better product came around it's market share diminished.
>>
>>109508857
>It seems like what is popular is what is predatory, has nothing to do with merit. More like vendor lock in.
Or simply not better?
Netscape Navigator didn't die because it was better, it was equally as shit as IE and people didn't feel the need to pay to use a barely better product.
It was not until genuinely objectively better products like Firefox or Chrome came around it began to change.
>>
>>109506893
If the code you're writing is in any way mission critical for anything, then yes, because if you use an LLM, you need to double check that the answer is correct by yourself anyway
>>
>>109508969
I addressed this here >>109508548
If you're writing mission critical software(or otherwise), you should be checking if the code is correct even if you didn't write it using AI.
>>
>>109508900
My intention isn't to argue with you, just to discuss.
There's a language barrier that I think is making our communication difficult, so I respectfully disengage. I have to write software anyway and I'm wasting time talking instead.
>>
>>109509028
You'll need to read the documentation anyway, so there is no time saved by using AI to look things up. That was my the point. Not reading the documentation can lead to subtle bugs, even if the code compiles, runs and seems to work.

>testing
>verifying output
Even if you devise the most rigid testing, you might still not be safe. Not all bugs can be caught in testing because not all bugs are deterministic (classical example is race conditions)
>>
>>109509179
>Read documentation
You don't need to. I've again addressed this here >>109508628
Also people are overrating the actual usefulness of most documentation these days as a reaction against AI. If documentation was actually that good, you wouldn't be having something like Stack Overflow. Programmers in general hate to write. As a result most of the time in actual software development, documentation is outdated or non existent anyways. AI is more helpful in that regard. You should be providing the library context to the coding tool, which most modern AI coding agents can do.
>Even if you devise the most rigid testing
Which is all the more reason for my point. Even you don't know what the code you've written does.
>>
>>109509266
>overrating the actual usefulness of most documentation
That depends heavily on the documentation and the library

Programmers might hate to write documentation, but hopefully they will have noted if, for example, a function modifies global state or has other important side effects
>>
>>109506756
May I ask what programming language the AI used? I'm kinda currious if it just wrote the 3 lines of bash that can acheive this or if it included a small LLM to verrify output
>>
>>109504918
I love snailcat pictures so I allow it
>>
>>109508628
>It eliminates the time I've to search, read, search, write boiler plate, test, rewrite boiler plate. The only thing I've to spent my time own is the final implementation of the business logic.
wow you must write some garbage jeet code if you're not unit testing. that's "saaar google play card i rape you tomorrow" level of dimwitted monkey-on-typewriter coding. Saars were doing that long before you or AI existed by just stealing shit from stackexchange. You're no better now than they were back then
>>
>>109509920
The documentation have only gotten worse because developers default assume you're going to use a ai coding tool to understand just like how mass adoption of IDEs drove them to stop explaining every function in the library and what they do.
>>109509973
Kid, I've emphasised the importance of TDD in this thread. Not going to repeat for a underaged edgelord again.
>>
>>109509953
i don't like pictures where snailcat is sad because i think about how much salt hurts them
>>
>>109505821
They aren't real programmers because they are illiterate, they are not trained to read code or even do basic math, they are posers. Even an UE4 blueprints script kiddie has better understanding than these guys probably, unless that is the target audience because they can at least do abstraction well enough to express what they want in vague code functions. Not that they can actually identify the functions they want to then use within the language of their task.

That's my impresison at least as a cider dropout.
>>
Here you go, I don't know how to write a single line of code and created this https://umigalaxy.com
>>
>>109504918
>ask vibetards to share what they have made with their supposed expert programmer machines
>don't bother looking at it because it's AI slop and you shouldn't waste time looking at something that didn't take any human-hours to create
heh, looks like nothing exists!
>>
>>109506322
Hello dumb fucking idiot.
>>
>>109506893
I used AI for search and eventually got burned with it wasting literal hours of my time hallucinating nonsense that sounded plausible, so don’t pretend like AI doesn’t also waste your fucking time.
>>
>>109508628
>calls others nodevs for not liking AI output
>thinks the quality of AI slop is good enough that it only needs a 10%-20% rewrite
lol
>>
>>109505821
They're a lot like trannies when you put it like that. Right down to the demoralization propaganda campaign
>>
>>109510480
>Kid, I've emphasised the importance of TDD in this thread. Not going to repeat for a underaged edgelord again.
you just keep reinforcing the monkey-on-typewriter stereotype lol
>>
>>109508485
>everyone I don't like is one person
Why are AI psychosed inbreds always like this?
>>
>>109511036
This but unironically
>>
>did yourself on 4chan to prove to us you did something
>no I won't stop making these threads afterwards

Op is a faggot
>>
>>109511932
>dox
The absolute state of vibetards. Embarrassing.
>>
>>109505821
They are genz script kiddies that's all
>>
>>109506018
this is the most indian coded repo i've ever seen
>>
>>109504918
>they never EVER share anything and just call you a tranny and post weird cat pictures
I've never seen anybody share their code on here
It's almost as if this board is full of poser faggots who spend more time talking about coding than actually coding.
>>
>>109504918
sharing results in copying.

screenshots, descriptions, can lead to indians copying.

Fact is, I only share with my gf, and I don't have a gf.

:^)
>>
File: Screenshot 17.jpg (107 KB, 1080x2131)
107 KB JPG
https://github.com/Titanean/Tadmor
>>
>>109512329
lmao
>>
File: ciabrainwash.jpg (113 KB, 1150x1208)
113 KB JPG
>>109513956
>It's almost as if this board is full of poser faggots who spend more time talking about coding than actually coding.
He learned guys. Arrest him.
>>
>>109513956
I've seen plenty of people share their code. But there is a reason people don't in general which is risk of doxxing. The thing is vibesloppers claim they can churn out tools for personal use (so likely not uploaded, easy to upload somewhere to show it off), or that they make millions with their vibeslop (could therefore easily point to it, we couldn't even tell if they're pointing to a random vibeslopped crapware that is not theirs anyway, the point from our side is only that we want to "see" this supposed successful vibegarbage that's totally good quality you guys it's not vibes it's AI engineering ok).
>>
File: bullshitmeter.jpg (47 KB, 600x602)
47 KB JPG
>>109514444
>I've seen plenty of people share their code. But there is a reason people don't in general which is risk of doxxing
>>
>>109514450
Take your meds, schizo
>>
>>109514461
They are just embarassed to share their vibe coded slop. The doxxing argument is complete bullshit.
Go back to /vcg/. Where retards like you belong.
>>
>>109514469
>illiterate zoomer can't read more than 1 sentence at a time
why are you zoomzooms all like that
>>
>>109506322
I'm torn if you're genuinely retarded or trolling.
>>
>>109514505
After seeing the first sentence, it was very easy to guess rest of the text would give any reader irrecoverable brain damage.
I'm so sorry you had to read/write the entire text. I hope your insurance company covers your diffuse axonal injury.
>>
>>109514513
>it was very easy to guess rest of the text saaar
>gets it wrong at the most basic level
Every time
>>
>>109514521
>Every time
Yes, every time. Every time you get called out for posting vibe-coded slop, you default to the exact same "/vcg/ zoomer/saar" script.
Now get to the hospital. They are waiting for you.
>>
>>109514535
Poo in loo zoomjeet, not in thread
>>
>>109514552
>ad hominem caused by being proven wrong 5 times on a streak
>>
>>109514561
This is what AI psychosis does to a pajeet huh
>>
File: donotdiffer.gif (1.42 MB, 1000x1000)
1.42 MB GIF
>>109514623
Take the L already. If there is anyone who has AI psychosis, its you for sharing your brain-damaged ideas and calling people insults because they didn't agree with you. You aren't coming back from none of that.
>>
>>109504918
I made a game which is a copy from one of those interactive mobile ads where you sort liquids.
Also I made solitaire.
>>
>>109509922
Sure, it used python, and tesseract and do not remember what for spelling and stuff
>>
I feel like the 3 groups that love vibe coding the most are
1. Grandprogrammers who have a shitload of experience, are tired of coding manually but still take the effort to review everything
2. zoomgrammers who never wrote a manual line of code in their life
3. third worlders
>>
>>109504918
Not a vibetranny but I vibecoded a 4chan archival thing today with dipsy, works well, I wish I could share it but you fellers will dox me.
>>
>>109505878
He's not the one sucking OpenAI dick and claiming "ermagehrd everyone is getting replaced by skynet" - you are. So you show everyone what amazing products you have built with it.
>>
>>109514646
Poo in loo, pajeet, not in thread
>>
>>109508315
>Why would I show you thing? You can see how many things I am making, I am lapping you every day of the week

1: You don't show the things. So no we can't see it.
2: No you're not.
>>
>>109515059
1. doesn't exist, because anyone with a modicum of experience knows it is far more annoying and takes far longer to review than to write the code.
>>
>>109515253
lol this.
It's easy to understand why vibecoderz can't comprehend this though. For them it's an obvious shortcut, for others it's a huge step back. The slopper is unable to see things from someone elses perspective and will instead seethe about actual programmers with experience for having a different opinion than them. And call them trannys.
>>
>>109515253
I dunno, some of the most enthusiastic I see about vibe coding are the senior guys in their 50s or 60s, including the open source space.
Then again they might have not coded anything themselves for over a decade at this point.
>>
>>109514646
>If there is anyone who has AI psychosis, its you
Do you have any idea how many thousands of times this is said between two or more people everyday now.

It goes
>Hey all that talking you to do that AI chat thing... I'm concerned for you. I've been seeing signs. There's this thing called AI psychosis and ... if you could just read this article please don't be offended
and the response is always
>Wow okay, if anyone here has "AI psychosis" it's you, you hate AI so much it has made you insane! Did the horseless carriage make you go crazy too? LUDDITE
And then someone has to restrain them
>>
>>109515373
I'll give you a hint...
They were always bad at writing code, so they literally can't tell that vibe coding is bad. Seniors are the guys who skillset is browning management and taking credit for other peoples works. Not writing code.
>>
>>109508466
>press button on machine to generate slop
>pay money to press the button
>have the machine glue slop onto other slop (if you are an advanced user)
>toss the slop aside and start your next slop
SEE? It's just a TOOL
>>
>>109511036
Show me the github repo of a successful vibe coded project made by a person/team with no coding/CS experience from the ground up. Projects like Linux and ffmpeg don't count because they've been worked on for decades, have thousands of eyes on them at all times and strict guard rails implemented by the tech leads.
>>
>>109504918
AI is fine for personal projects you want to roll out in a day and don't expect anyone to use.
>>
>>109515478
This. AI is only good for things you conveniently would never want anyone else to see or use. But also you will be left behind for not also making forgettable throwaway slop using AI... or at least that's the narrative for the past 2 years?
>>
So what is AI good for now? Tinkering.
And who loves it? Tinkerers.
Also slop pushers who have flooded the internet to the point that youtube is now removing all AI slop and slop channels and slop bots from their platform.

It's pretty good for that.
>>
>>109513989
>actually a cool thing in the thread
>no reply or quote from neither side of the people
nta but i guess people here are arguing for the sake of itself
>>
>>109515373
Yeah, those are the people who haven't written code in 40 years and whose idea of extremely complex programming is writing 10 lines of C that doesn't segfault. They are for all intents and purposes nocoders in practice. They don't review any of the code they output, at all.
>>
>>109515465
https://github.com/odysseus-dev/odysseus
>>
>>109504918
I'm pretty close to releasing my vibe coded DAW/tracker. I've been working on it for 3 months and it already surpasses full price software in some ways
>>
>>109516348
kek
>>
I made money. I'm not sharing it. End of discussion.
>>
>>109516348
that issues section jesus lol
>>
>>109516365
How?
>>
>>109504918
It's about making something. It's about sending a message to the leftie luddites trying to hold back progress because they're still mindbroken by Trump and Elon.
>>
>>109506597

you get free tokens if you are influencer
>>
>>109517627
Not going to give away all my secrets, but imagine a tracker with the sequencing flexibility of a modern DAW. Each clip on the timeline can be a different sequence of a module's patterns. Each channel is a separate module with its own clock. In short, you are not bound to a single master grid, although can be if you want to.
The hard part so far has been compatibility with existing formats.
>>
>>109505027
>11 days and $25
>to overlay 2 GIFs
Feeling scammed yet?
>>
>>109504918
>I-I-I'm n-n-n-nnot w-w-w-worried!
>makes the same threads 20 times each day
Yeah we can tell how relaxed you are.
>>
File: roboloid.png (405 KB, 1603x989)
405 KB PNG
>>109504918
i made a earthsiege model importer, it properly assembles the robots from their data pose. currently refining the animation import process. i don't know how to screencap a 4chan webm easily to show the animation playback in blender. maybe i'll vibecode that on the side.
>>
>>109515373
it's great for comprehension on projects written by fortran mummies, quick fixes on legacy codebases, and unspaghettifying old hacked together bandaid projects that both under- and upper- levels wont give you enough time to go back and fix manually.
t. business basic grandwizard
>>
>>109514513
>>109514521
Is this the absolute state of snail cats? Lmao.
>>
>>109511227
I've not called anyone nodev for not liking AI output. Please learn to read. I said if you're someone who's main argument against AI is that you've to READ what they wrote, then you probably weren't doing much work anyways because actual devs read and verify everything they write, even hand written by themselves. Most critical hacks happened because of an arrogant dev that pushed his code without reading and just guessing it'll be correct regardless.
>>
>>109511748
Sure, whatever cope that helps to float your boat.
>>109511819
In case of that post, it's obvious from the same kind of posts, it's the same guy. Of course, I know there are indeed many retards like you.
>>
>>109511169
No you didn't, you just repeated whatever some youtube retard told you lol.
>>
>>109515394
I feel like /g/ is filled with failed retards who would be better off going for trades instead of seriously deluding themselves like this.
>>
File: open-india.jpg (159 KB, 1168x735)
159 KB JPG
>>109519462
(nta)
sure, theyre so superior in writing code that openai, with their frontier models and infinite tokens, rather hire indians than to trust their ai
>>
>>109519510
Indians are cheaper than tokens.
Anyways that's not my point. /g/tards have been plagued with /pol/tard and reddit disease of people who I don't like = retard.
You're seriously delusional if you think John Carmack is a worse programmer than you just because he's pro AI.
What's funny is, I've never seen anyone post this so called mission critical software or any of that shit from /g/. Even a /v/tard has successfully made a billion dollar franchise. It's usually just some kernel monkey who's just a drop in the ocean of kernel monkeys.
>>
>>109519581
>indians are cheaper than electricity since its openai themselves were talking about
anal retardation.
>>
>>109519581
>>109519621
tfu
>more expensive
that was so retarded it destabilized me



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.