[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / r / 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

Name
Options
Comment
Verification
4chan Pass users can bypass this verification. [Learn More] [Login]
File
  • Please read the Rules and FAQ before posting.
  • You may highlight syntax and preserve whitespace by using [code] tags.

08/21/20New boards added: /vrpg/, /vmg/, /vst/ and /vm/
05/04/17New trial board added: /bant/ - International/Random
10/04/16New board for 4chan Pass users: /vip/ - Very Important Posts
[Hide] [Show All]


[Advertise on 4chan]


File: file.png (85 KB, 716x514)
85 KB PNG
RPCS3 devs telling it like it is.
>>
>>108801411
Preach sister!!
>>
>>108801411
Codetrans Luddites are obsolete and refuse to accept that they have been superseded by VibeGods. No one writes artisanal code anymore, everyone is just using AI to boost their productivity.
>>
I don't like normalfags using the word slop. It was MY word!
>>
main issue is people generating ai code and not providing documentation or actual "why this shit needs to exist". human does things for a reason. ai does things because ???
>>
>>108801411
based
where's snailcat poster now?
>>
>>108801712
lol closet homo.
>>
>>108801411
I get his point but it's ultimately pointless, there's no stopping it. Not that I'm defending broken code, but what can he do about it? What's stopping people from refactoring code ad aeternum until they get what they want from it?
>>
>>108803133
They're prohibiting AI slop, not AI code.
>>
>>108801411
prompstitutes BTFO
>>
>>108803133
its pretty simple. code cannot be accepted unless they actually write out a usecase and proof of work. like your math teacher telling you to show your work on paper. cheaters/ai cannot do this without going into ai slop territory.
>>
File: agdgpill.jpg (664 KB, 1448x1086)
664 KB JPG
why are they so afraid of ai? these people are ridiculous.
>>
>>108803143
>They're prohibiting AI slop
And how will he enforce it? What constitutes slop? What makes "AI code" different enough for it to be accepted there? What will stop people from just making their own fork off the main project and let anyone who wants to brute force it into something functional via vibe coding? You see where I'm getting at? This all pointless drama.
>>
>>108801712

Dilate skillet ^^
>>
>>108803173
Humans write code to fix a problem they want to solve while preventing regressions. AI writes code without any context.

Ask a math pro to show your work on paper and he will do it. Ask a cheater to show your work on paper and they stall and hallucinate answers.
>>
>>108803198
That doesn't answer a single question I've asked, it's just silly anti-AI bias. OK, you hate AI, what now? There's nothing to it. It sounds the same as being vegan or some climate change activist. It's all pointless shit with no impact whatsoever.
>>
>>108803169
>(Actually
>(Finished)
>>
>>108803212
calling someone a vegan is the most midwit insult of all time
>>
>>108803212
The fact you used AI doesn’t give you the special right to spam nonesense code at maintainers.
>>
>>108803133
>hat's stopping people from refactoring code ad aeternum until they get what they want from it?
Nothing except their own laziness.

If you want your AI code to be inside of RPCS3, then what you have to do is
1, get your AI to fork RPCS3 for you,
2, get your AI to implement new features in your RPCS3 fork,
3, advertise your fork and convince users to try out its features,
4, build a strong enough fanbase for RPCS3 devs to take notice,
and 5, wait for RPCS3 devs to absorb the code whose value you've proven in the real world by getting users to willingly and knowingly use it.

Get forking.
>>
File: 1778540947571817.jpg (764 KB, 2905x547)
764 KB JPG
>>108803169
i spent less time, you defeated
>>
>>108801795
every time i go outside i see a muni bus with an anti ai slop ad. i hate you because i have to be reminded of you every time i step outside
>>
>>108801411
We need a list of AI-free software like we have for opensource software on github.
Anyone published such a list yet?
I cant find anything because poogle interprets "AI-free" as "Free AI"
>>
>>108803173
ai slop = it doesnt compile and run
>>
>>108803177
Are you blind? lowqualitier
>>
>>108803173
Slop produces text "artifacting/model bias" that can be fingerprinted similar to a browser fingerprint
>>
>>108803173
If humans are reviewing and debating additions or changes then there is a manpower limit to how many commits they can actually make. It doesn't matter what is causing an increased volume of changes, if they can't sort through it all it's not going through. Brute forcing it just means lowering standards for what goes through.
>fork project
>train AI on project
>hand every change pushed to AI
>if it approves add it
>no human meddling maximizes throughput
Tell me why we aren't doing this for every piece of software on the planet without sounding like a luddite.
>>
it's the equivalent of jeets wanting to get a pointless rewording of some documentation merged so they can say they are Certified Open Source Developer who do the needfuls
>>
>>108801773
LMAO. Ok, Brainlet: https://www.youtube.com/live/zwYkHS8jvSE?si=T1F3Vd4n5IMyjIhM&t=4840
>>
It's funny how obvious it is when they pay the glavset to defend something, look at all the garbage damage control on the thread.
>>
>>108801411
Snailcat rage lolololol
>>
>>108801411
based wtf I love rpcs3 now!
>>
>>108803173
>And how will he enforce it? What constitutes slop?
Went over this in the last thread about this dumb shit with autists pretending to be retarded.

1. Mods know it's AI slop
2. You know it's AI slop
3. You aren't fooling anyone

This isn't a legal contract, there isn't a court of law with a judge and lawyers and a jury rendering verdicts. No one's going to jail. If these OSS jannies flag a few false positives and people get warnings and bans they technically shouldn't have had, it's no big deal.
>>
>>108804225
I doubt anyone will bother with this.
It's like saying you could use an AI model to identify someone being faggot or a troll.
>Cool Guy starts a hobby forum
>Establishes Rule #1: don't be a faggot
>You register and immediately begin posting faggotry
>Get warned
>You cry about it, claiming mods are wrong and your posts technically aren't faggotry
>Get summarily banned because you aren't wasting any more of their time
It's the same thing.
>>
File: 1764198217130210.jpg (204 KB, 900x1200)
204 KB JPG
>be me
>start vibecoding at work
>annoying copypasting shit into it
>idea.winamp
>give access to entire repo and passwords etc
>start off small, this is great, "ai please commit the changes"
>suddently, big job comes in, quite complex changes
>no worries i got this.avi
>boss super impressed
>type in big prompt what i want
>it thinks for about 10 minutes
>code looks a little weird but wtf do i know its ai so it writes perfect code
>"ai please commit"
>some permissions error because it needs to commit to new repo i dont have access to
>wtf is that for, no idea just need to push these changes so i can impress boss and get promoted
>i tell it to just upload the code ffs
>it somehow found some system git user and committed a bunch of garbled bullshit to the main repo, publishing the changes in teh meantime
>company goes down
>everyone screaming
>trying to trace wtf happened
>wtf is "SYSTEM" user?
>boss finds out code is written by my friend claude
>i get fired
>i angrily tell claude it fucked up,
>claude said good catch and that i am very smart
>>
>>108801773
Peter Theil shill in the thread
>>
File: 1760189066377682.png (514 KB, 880x898)
514 KB PNG
>>108801411
Emulator developers are always in meltdown mode, on top of having to tell kids to dump their own games now they have to handle the slop of AI bros. they deserve this for harassing me almost 20 years ago
>>
>>108803169
If writing code is the same to you as being in prison, there might be a deeper issue.
>>
>>108801411
Good to hear they have standards for contributions
>>
>>108801773
this project is an emulator, not your retard-proof frontend web development job
>>
>>108808114
I doubt you were ever important enough for people with skills to harass you
>>
>>108808079
As long as we cannot horrifically torture LLMs to make them actually give a shit they are completely fucking useless. They NEED to be able to perceive pain and fear.
>>
File: 1759653847138414.png (298 KB, 1044x1582)
298 KB PNG
Unsurprisingly, retards ITT don't care about AI or code and only treat it as another culture war to participate in to kill any and all rational discussion.

RPCS3's guidelines are perfectly rational. They're not anti-AI, and embrace the fact that it has it's uses, but they're against talentless hacks that use AI as a replacement of skill that's needed in something as complex as a PS3 emulator. They're 100% on their stance and anyone seething over it has no right to work in IT.
>>
File: 1754367096086967.png (462 KB, 564x852)
462 KB PNG
>>108808415
I literally got harassed by two emulator developers in 2007, it didn't have to do anything with emulators. I'm glad when bad things happen to emulator developers and I want all the AI bros to invade their circles
>>
>>108808650
>I have a 19 year old grudge and I want all emulator devs to suffer because of it
You sound as insufferable as emulator autists.
>>
>>108808650
can i get a qrd?
>>
File: 1765854793687467.png (943 KB, 968x781)
943 KB PNG
>>108808682
>>108808685
don't wanna dox myself but i basically got harassed by 2 people back then, some time later i figured out that they were well known in the "emulation scene"
ever since then i wanted them to suffer and they've been suffering a lot lately
>all the android third worlders
>people constantly asking where to download games
>now all the AI bros invading them
they deserve this
>>
>>108808685
qrd
>>
>>108808650
This never happened.
>>
File: 1766706614819193.png (299 KB, 443x547)
299 KB PNG
>>108808792
my bullying happened in 2007, I don't know if they were a part of the "emulator scene" then but a few years later around 2010 i found out that they were well known in that "scene" by that time
>>
>>108801411
BRVTAL
>>
>>108801773
retard
>>
>>108808603
I would consider myself somewhat of a "real" programmer, and even I can freely admit i've used AI generated code that I did not fully understand.

The difference is I ran it locally for 45 days, found a few edge case bugs, fixed those, ran for another 30 days or so before finally deciding to deploy it.

I still don't know exactly what it does, but it DOES work.
>>
File: pigreallylikeslop.jpg (32 KB, 371x490)
32 KB JPG
>artisanal code
The thing that the shills don't get is that the marginal costs of software are trivial. It doesn't matter how much slop vibecoders produce, if none of it is better than the best artisanal competitor, it may as well not exist at all. It's not like manufacturing where mass-produced products defeat the hand-crafted competitors through massive volume.
>>
>>108801773
they're quite literally allowing ai shillsis, as long as it works
>>
>>108809571
if peeps weren't delulu about LLMs they would realize they are incapable of reasoning
>>
>>108808603
it's not even against talentless hacks, it's against jeet tier people shitting all over the repo with ai hallucinations
>>
>>108809571
>midwit take
AI is crushing your artisanal code because it's x100 faster and deliver x10 your features. Here's your massive volume difference dumbass
>>
>>108809693
In a buggy, non-performant way. It's literal white noise.
>>
>>108801795
>"Slop" has evolved from a 15th-century term for "muddy water" and a 14th-century term for "loose clothing" to mean low-quality, mass-produced content.
>>
>>108801795
I saw an ad for “___-maxxing” in the subway, it’s over
>>
File: 1747284398536470.gif (527 KB, 220x187)
527 KB GIF
>>108801411
oh good I was worried the A.I coders would fuck up all the hard work emulator bros have made over the past 20 years.
>>
>>108801411
open source troons are deranged. leave something useful to humanity? you mean feed the ai and all the jeets at its teat?
>>
>>108808727
why did they hairass you? You weren't using RPCS3 to cheat at speedruns were you? I'm telling Karl Jobst
>>
File: bullshit.png (83 KB, 570x558)
83 KB PNG
>>108808650
>>108808727
>I dindu nothing and I was literally raped
I'm noticing the pattern
>>
>>108808186
not just writing code but development of games where you never finish a game.
>>
>>108803133
>what can he do about it?
Bully, humiliate and gatekeep.

On a wider note, it's a broader concern that automated slopspam can be weaponized to cripple open source.
>>
>>108801411
>hey if I call it slop I won't have to actually articulate anything wrong with it.
>>
>>108803292
Because its actually impossible to know if it was generated with ai. Same reason they can't just block or ban bots.
>but it sucks!
So does most human written code.
>>
File: 1778498602954314m.jpg (157 KB, 997x1024)
157 KB JPG
>ethics in playing games for free on your computer
I don't get it, it's like a fentanyl dealer refusing to deal to bigots
>>
>>108808727
>troonime avatarfag is mentally ill
big surprise
you deserve to get bullied
>>
>>108803173
>And how will he enforce it?
they decide what it is and how to enforce it, because it's THEIR PROJECT. is that too difficult to understand? do you even program, retard?
>>
File: snailcat-vs-aichads.jpg (1.29 MB, 3464x3464)
1.29 MB JPG
>>108801773
Luddites can only cope and sneed
>>
>>108811356
Yeah that's how it works when you have both credibility and authority; or when the flaws in the offending material is self-evident to everyone who matters (Not You).
>>
>>108809693
>seething and lying
Good devs are using AI to make good code faster, it's true.
But no, the kind of high-volume, vibecoded slop that's being banned by RCPS hasn't defeated artisanal code at anything. The only place vibeslop is successful at all are new domains where nothing yet exists at all (meaning that as soon as someone good actually programs a real solutionk, the vibeshit will die)
>>
It must genuinely be hell to have to look at the code that retarded Indians who can't even prompt correctly because they can't even speak English properly never mind code, are submitting to projects.
>>
>>108803167
>like your math teacher telling you to show your work on paper
Kinda off-topic but you wouldn't believe how hard it is to get kids to do this, even in college. Any time you give them a problem that requires any more explanation than just circling a number and you get back a complete garbled mess of half remembered theorems and definitions. I can't blame them too much though. There's a systemic issue in that our math curricula are designed to create mindless calculators rather than develop actual critical thought
>>
>>108809571
>Moore's Law totally isn't dead
Holy midwit.
>>
>>108808650
>>108808727
>>108808832
It's been nearly 2 decades mate move on
>>
>>108811851
And in all likelihood it probably was something completely benign that he misconstrued as harassment because he's mentally ill
>>
>>108811422
there's literally not a single reference to ethics
>>
>>108811669
>it's true.
Evidence?
I have never seen a good dev go anywhere near AI because the unanimous opinion is it can't even do intern-level code and constantly gets even the simplest things wrong, thus being 10x slower in practice. It's also what the research on AI productivity shows (generally, people report a 20% boost in perceived productivity on average, but the measured productivity is in fact 20% down -- a 40% perception gap).
>>
File: 1768467310435498.png (616 KB, 873x1090)
616 KB PNG
>>108810464
>>108810718
>>108811438
>>108811851
at the end of the day its only gonna get worse for them
i was laughing my ass off over the years over all the indians and pablos telling them to kill themselves because the emulator doesnt work on their fisher price phone from 2013 or asking where to download mario whatever game, that always started a mental breakdown in their head
now they have to deal with all the retarded ai bros
you deserve this for being a faggot piece of shit all those years ago
>>
>>108808832
>>108808727
>>108808650
you seem like a massive faggot. I believe they were fully justified in their actions regardless of what those were. you brought it upon yourself.
>>
>>108812223
You sound like it would be great fun to torture you.
>>
>>108812223
You are wallowing in your own sorrow for the rest of your life over mean words from the 2000s.
Grow up you sad pathetic faggot
>>
>>108812202
>the unanimous opinion is it can't even do intern-level code
unanimous by whom? claude can easily outperform any new grad in code generation, the thing it fucks up at is understanding massive codebases and the hidden constraints in contributing to them, because it has a limited working memory.

but even then it makes better assumptions than most humans. on a large project you have to check and guide its work but the same is for a human junior dev
>>
>>108812278
>claude can easily outperform any new grad in code generation
Stopped reading there.
>>
>>108812278
>unanimous by whom?
Anyone I talked to in a professional context whose code was able to pass peer review lately.

>claude can easily outperform any new grad in code generation
Maybe if your idea of new grad is yesteryear's idea of a newborn. I will give you that people's idea of interns and new grads has always been far, far, far below the level I have always been held to in my professional life from day 1 and I've never seen someone perform that poorly not be fired on the spot (typically those weren't bad people, but people pretending to work instead of working), so maybe my impression of new grads is skewed. But those are "so bad it's better to have nothing" tiers of people, so you're really not making an argument in favor of claude here.
>>
>>108802286
That's a big part of the issue indeed.

But the problem is that a lot of people, in particular jeets, are using AI to quickly generate code for pull/merge requests for OSS projects so that they can build up their profile for job searching purposes. They already did that before AI by just refactoring variables of messing with readme files or doing other useless shit, but now that they can pretend to code its a serious issue.

AI code by itself can be useful but it needs to be verified, debugged and validated and also conform with the any guidelines and ideally solve a published issue. Most AI code submitted is just random garbage that may or may not work.
>>
>>108812307
Plus, with pervasive CI/CD pipelines, it wastes metric fucktons of resources on automated testing that will just flag everything as broken.
>>
>>108812302
>I've never seen someone perform that poorly not be fired on the spot (typically those weren't bad people, but people pretending to work instead of working)
I see retards get hired and survive all the time. Well, less so nowadays because nobody's getting hired but certainly 5-10 years ago.
>>
>>108812284
maybe if you only hang out around 160IQ MIT grads. the average codemonkey is a fucking mouth breather and always has been.
>>
>>108812447
I guess I just keep getting rejected when I apply to these companies then. They probably specifically select for retards. I mean when I was at m$, I wasn't exactly surrounded by the brightest bulbs in the shed but that was nowhere near AI-bad.
>>
>>108812458
Guess I'm a 160IQ MIT grad then.
>>
>>108811338
>Bully, humiliate and gatekeep.
gatekeeping an open source project?
>bully, humiliate
touch grass please
>>
>>108812202
>I have never seen a good dev go anywhere near AI because the unanimous opinion is it can't even do intern-level code and constantly gets even the simplest things wrong, thus being 10x slower in practice.
It can be used well to solve certain classes of problems exceptionally well. For example, suppose you have like 50 class or type definitions that follow a pattern. You want to update them, but even with IDE refactoring tools it would be a huge pain. You can tell AI: look at this class, then change this other 50 classes in the same way. You can fire off an agent with a reasonable hope that it'll give good results, and let you do something you probably wouldn't have even bothered with before AI.

It can also be used effectively to quickly gain understanding of unfamiliar systems and codebases (dependencies, platforms, etc). You can dive into lower level systems and figure things out that would have taken 10x as long before AI. You can use it to translate logic and algorithms from one language to another with pretty high success. It can be very good at spotting bugs and mistakes. It's very good at finding security vulnerabilities (at least, well-known classes of vulnerability likely to appear in your codebase).

AI is also good for generating "throwaway" code and personal tooling. Most devs will often employ scripts and small programs that aren't destined for production, they're just for testing an idea or automating a tedious task. AI can be really helpful in speeding up that kind of work, allowing devs to create more and better personal tooling.
>>
>>108812278
>claude can easily outperform any new grad in code generation
NTA but I probably wouldn't keep reading either, it's a pretty dumb comment.
1. "New Grad" is a terrible standard. Code from fresh grads has been known to be worthless for decades.
2. "code generation" is a terrible metric. Nobody with a brain pays programmers to merely "generate code," they pay them to solve problems. Code is just the medium.
>>
>>108813105
>gatekeeping an open source project?
Open source doesn't mean everything gets approved, fatchild. It means you can fork using your slop commits and good luck with that shit.
>>
>seething

When did this word become common to use
>>
>>108811834
Moore's law never applied to that point one way or the other, retard.
>>
>>108813888
Zero sum game where the only winning condition is to make other people uncomfortable.
>>
>>108813937
>the midwit actually believes this
>>
>>108813787
>but even with IDE refactoring tools it would be a huge pain.
Why? Be concrete.

>You can tell AI: look at this class, then change this other 50 classes in the same way.
That's just about everyone's first ever test for AI. This or similarly easy tests like 'implement this 5 lines pseudocode in java' or whatever the fuck. The problem is precisely that this is where AI works worst of all. It fails nearly 100% of the time on these supposedly obvious wins for it. That's also why people are more skeptical of it than they ought to be.

>It can also be used effectively to quickly gain understanding of unfamiliar systems and codebases (dependencies, platforms, etc). You can dive into lower level systems and figure things out that would have taken 10x as long before AI.
I've only ever found the opposite to be true. It keeps inventing methods that don't exist, invents meaning to methods that have nothing to do with what they do, invents the non-existence of methods, etc. even in well-known libraries when provided direct access to their docs, let alone something internal.

Sounds like you never actually used AI for anything and you are just ideating as to where it could be useful.
>>
>>108802286
Figuratively this.
You can gen borderline perfect code (I mean, (You) can't, but it's possible) and the second someone needs to modify it, debug it, or expand upon it, good fucking luck.
Properly coded and documented software is still going to reign as winner in the slopwars. Think otherwise? Go pull some randoms' AI-gen code from github and try and fix the problems for them.



[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.