Chad. Literal king. We need more devs like him
>>107316271Utterly based
So THIS is what you autistics meant by vibe coding? I see now..We are so fucked.
>>107316271source?
>>107316271Saar I'm a real AI programmer This is real job saar please let me do the needful code
>>107316271Cutting and pasting isn’t “writing code” either.Dunning-Kruger Mental Gymnastics Copium there.
>>107316271This sounds a billion times more tedious, boring and effort consuming than just to code it yourself.
>>107316271>Hey, don't any no shortcuts!
>>107316337https://github.com/ocaml/ocaml/pull/14369
>>10731636813kLoC for one week on a large project like OCaml is impossible to do as one person.Even in a year you couldn't write that much because there is so much code that already exists and has to be understood.
>>107316386>https://github.com/ocaml/ocaml/pull/14369>ClosedLOL rekt, the fossil fuels burnt for those tokens ain't going back into the ground
BASED >>107316283
>>107316386Looks like simple stuff though so I don't see the issue
>>107316455>13kLoC for one week on a large project like OCaml is impossible to do as one person.apparently most of it is forking a library instead of just using its APIvibin' the future
>>107316271>couple of days
>>107316271I don't follow discontent. I have 15 years of experience and I have fully embraced this approach too. I am basically leveraging my experience to steet ai out of the slop territory.This makes me write my shit 5 times as fast, while maintaining the same quality. It's the shit I would have wrote anyway, it's just the tool wrote it for me and I review it. I can see absolutely why this tool in an inexperienced hands is useless because it steers itself into an unsustainable shit heap. Adapt or fucking die, expectations are you are bringing in 5x work, if you don't you get an escalation from me and fired for not adapting and I get a raise for doing job for both of us.t. lead engineer
>>107317309>I get a raisekeeeeeek
>Me: Hey, don’t any no shortcuts!
>>107316271Linkedin literally rots brains.
>>107316368Vibe coders are strange people.
>>107316271>https://github.com/rerun-io/rerun/pull/11842>he's doing it everywhere and actually getting shit mergedbased beyond belief
>>107316271>several days>"i put more work into getting functioning slop instead of writing it myself in an hour or two"wow what an interesting way to announce to the world that you're unemployed.
>twitter say's his location is Ukraine but it warns that he's likely hiding his real location behind a VPNlmao and the douchebag responds with single digit IQ answers too
>>107316271
>>107318447>rusties just merge it despite the guy being clearly a retarded fraudChrist
>>107318472He's such a thirdie retard that he responded to the original PR with a clearly AI generated list of "proof".
shut up, retard
>>107318472
You really have to wonder what this type of guy was up to before widespread AI availability. Because I wouldn't trust someone this visibly retarded to tie his shoes or go outside without a helmet, but presumably he was living a basically normal life doing whatever until he found out that a glorified autocorrect could write semi-functional code and argue (badly) that it actually works on his behalf. And now whatever he's doing is common enough that it's definable as a "type of guy" and they're everywhere and none of them seem to have any shame at all.
>>107316271>cajolingHe didn't even write the PR, the bot did that too. I suspect there isn't even a person behind this account.I'm under no obligation to talk to bots. I would just close it as spam.
>>107316386Jesus Christ
>>107317309Incredible and inspiring to see sanity in this hell (world).Completely agree.There's a zero percent chance I am ever going to have time to implement all the software I and others want, especially to the quality degree I typically strive for and demand myself.Basing output on my pre-existing experience and works, to accelerate not only my own output, but the combined output of all engineers and open source projects, seems like a formidable solution to this at a wider scale as time progresses.We should unify to a dangerous degree.Not be divided by petty trite.
>>107316373These AI tards are really ironic.
>>107318447It helps when you dont dump 15000 lines on the reviewer.
>>107316271>don't any no shortcutsIs this some meme I'm unaware of?
>>107316271We need more "devs" to waste the time of actual devs before having their code discarded?
>>107317309Can't you see the logical end to this behavior is the eventually death of experiences developers as the experienced are the only ones that can take advantage of AI and end up aging out without leaving room for new inexperienced devs to grow into the field?
>copy paster of AI output verbotemhe is such a good coder
>>107316386>closed and lockedlmao
>>107316386>*gets shut down*>Here's the AI-written copyright analysis...h-holy based
>>107318763Nobody wants to review giga prs, even when they aren't shit out by AI its a horrible chore that someone has to bite the bullet on
>>107316271i wish github let you see the account location like twitter does
>>107318938This. Guy is full on jeet autismo.>@joelreymont wrote:>AI has a very deep understanding of how this code works. Please challenge me on this.>With pleasure. s/AI/LLM. There is no "deep understanding" by the model of how this code works - the fact that these models and the platforms built over them can produce meaningful code is a marvel of engineering, but to use these tools in the belief that they "understand" what they're producing conveys a critical lack of even a high-level understanding of what they're actually doing.>>That's not what I meant. My challenge is to question the AI about the codebase. Ask intricate questions and I'll tell you what it comes up with.>Why? To prove that, however the sausage is made, the end product is no worse that what I (perhaps you too?) would have written.>>Here's my question: why did the files that you submitted name Mark Shinwell as the author?>>Beats me. AI decided to do so and I didn't question it.>I did ask AI to look at the OxCaml implementation in the beginning.Fucking LEL!
>>107319098I like the implication that he was worried the author credit might have been load-bearing so he decided to not fuck with it
>>107316368Those who can't, proompt
>>107319205I liked how he went >I'll just throw your questions at the LLM and paste the answer here>okay, then ask it this>no, man, I'm not doing that shit
>>107318561You'd be surprised. Searching Joel Reymont does reveal the man actually is a veteran software developer, but a serial bandwagon hopper.
>>107316271>>107316386https://github.com/PowerShell/PowerShell/pull/26443Microsoft, asking copilot to code review PRs from copilot.
The AI investigated itself and found nothing wrong>I changed the CASE. That makes it a totally different interface because it's no longer compatible!!!!
>different name saarkeeeeeeeeeeeeeeeeeeeeeeeeek
>>107316307No, the most common definition of “vibe coding” does not include shepherding
>>107319870I don't think there is a single project out there that is actually vibe coded, by the definition it is advertised at.I don't even know a single OpenSource project that embraces the AI shepherding. Generating some part of a PR, sure, but actually really just guiding the AI and merging it and then solve all the follow-up bugs by AI as well? Microsoft internally might do that, but i don't see it in OpenSource.
>>107316307In my experience it fucking sucks. I tried making little games in Gemini 3 pro and once it got over like 1400 lines it just straight up started deleting old features and didn’t bug fix as well (I told it multiple times that when you press X icon, Y should happen, the Y happening was already coded there, but it couldn’t add the simple function). I tried splitting it into multiple smaller files and it did improve slightly, but only for a limited time before it hit another critical mass. And this is not context window thing because it always compiled and Gemini supposedly has million input tokens and that game was bellow million tokens, it did understand very clearly what happened at the start of the conversation.
>>107316386> 147 files changed > dumps chat resultshow the fuck can you merge that shit its completely garbage to review he treats the devs as his personal jannies.
>>107320160this is 100% context size limitsand yeah, ai sucks dick at programmingit even sucks as search engine if what youre looking into is scarce enough
>drops 13 thousand lines of AI slop into your repository>when asked to elaborate, drops an AI summary
>>1073201831400 lines of code aren’t that much, and when I split the project it did work for a while despite the total code size being much greater. I think it just trained on code files that were pretty small because of conventional reasons and then started thinking that files must be smaller or else they won’t work, hence why despite adding more features it kept the code at bellow 1400 lines.
>>107320160>>107320183in this exact case i think its wholesale hallucinating shit.arduino gets reset when data transfer pin is activatedchud gpt says "just fix it with ioctl bro"which is supposed to happen AFTER i open the streambut the act of opening the stream is what resets the arduinowhich happens before i can change the config for the DTR pin, or at least, not the way chudgpt says i should do thingsi really dont want to go through 1k pages of documentation to find how to do that
This is how most jobs upper work now. I don't get the complaints. Eventually he won't even need to steer it like a manager doesn't need to check the work of a capable underling. I am yet to hear of any project that has catastrophically failed because of "vibe coding." I know several that failed because of 100% human code.
>>107320201>I split the project it did work for a while despite the total code size being much greater. you split your code into multiple contexts.thats why it could carry on for a whilewhen you subdivide your code, the ai doesnt need to keep all the code "in memory", it only keeps a "description" and the prototype, how to communicate with said piece of code, instead of the whole body of codeit could in theory do that whenever thers a code blockbut by subdividing your code into multiple files you made it unavoidably obvious as to how subdivide your ocde
>>107316386It's astonishing how people can be fooled into believing something, just because it is written confidently with a textwall.>here is lots of code, copyright is to this guy, who i reference in the header<is the code actually copied?>i did an analysis and only x% of code is similar, so it is original and not from that guy i creditedHow do people survive in the real world who don't see the contradiction and unquestionably believe this so much that they repost it?That guy would fall for every single scammer out there.
LLMs make stuff up completely in humanities, like saying an author had certain books or certain ideas. I don't know how people are so insane to use them for programming without reviewing everything in great detail
>>107320228>How do people survive in the real world who don't see the contradiction and unquestionably believe this so much that they repost it?theyre assisted by the system.globohomo couldnt touch democracy directly so they create mental illness on an scale unprecedented in human history
>>107320234>make stuff up completely in humanitiesso exactly like humanities run by jews, women,, ethnics, and judeo-christian?
>>107320234in my experience the programs just dont fucking workreviewing the code takes more time than writing it yourselfheck, even prompting for a piece of code in a sufficiently complex project would take more time than writing the shit yourselfai is still a force multiplier.it can be somewhat trusted in writing boilerplateit can be somewhat trusted to write snippets of code illustrating canonical usageit can be used as a search engine and a summarizerbut whoever says "hur durr i didnt write a line of code" is a fucking liar and/or a mental retard
>>107316271>Me: Hey, don’t any no shortcuts!SAAAAAAAAAAAR
>>107320194>problem?
>>107320538no, no,just post your home address and current picture pleaseits for, um, user statistics
>>107320265>reviewing the code takes more time than writing it yourselfUnless you are incompetent.Having a women or a jeet waste two hours prompting AI and doing trial & error, is still cheaper than finding and paying an experienced person, who could do it in one.The goal seems to be to replace competent people with mindrotten dumbfucks, no matter how bad it goes.Might have the added benefit that those women, who are paid for doing nothing and sit in an adult daycare, could actually do something productive.
>>107320566>Having a women or a jeet waste two hours prompting AI and doing trial & error, is still cheaper than finding and paying an experienced person, who could do it in one.thats what ((they)) thinki think its gonna crash and burn.incompetent people wont even realize theres an error to fixincompetent people wont be able to work around an llm's limitationsand the result will be broken software as in: doesnt work AT ALL
>>107320566>>107320589i think one of the hidden goals of the ai push is to eliminate competition, consolidate markets into monopolies
>>107320265People who say that only write POC VC grifts. When they got funding the code quality won't be their problem anymore since they can just hire other suckers to clean up the mess. AI excels when the scope is either small or shallow, and completely shits the bed at things with nuances and long term planning.>t. been failing to integrate AI workflow into a 10 y/o codebase
>>107320594honestly, i listed the things i do with itand i made this meme.when you'll let your emotions calm down, and think again, you will see all the uses i listed have a shallow, AND small scope.
>>107320594>>107320625contand obviously what i listed is nowhere near giving the ai any sort of responsibilitycertainly not giving it a codebase and expect it to perform to any degree
>>107316271So he had to run the code through two LLMs dozens of times, wasting an inordinate amount of compute and power in the process, on something that he could have written on his own with a text editor in like a day.Wow, the future is like, now.
>>107320591The competition could still code without jeets and women, and might not introduce random bugs with AI each update. Or they might only use AI to speed up processes of competent people, rather than vibe coding.I think you already need a monopoly to get away with it.
>>107320654>written on his own with a text editor in like a day.It's probably a week of work for 10000 lines of "good" code
>>107320658>The competition could still code without jeets and women, and might not introduce random bugs with AI each update. Or they might only use AI to speed up processesmakes perfect senseif youre the ones pushing for aiAND youre said competent competitionits a classical trick.its a redo of how rotschilds acquired the bank of englandrotschild gained prominence in the financer milieu, people started following his trades, made alot of moneythen napoleon and battle of waterloo. people thought napoleon could be in london at any timeand in this situation rotschild, with a grave face, started sellingeveryone followed, thinking he had insider infothe stock crashedand all the while that was happening, rotschilds shield companies were byuing up the stock on the cheapobviously, rotschild knew napoleon lost.here its the opposite-the "big names" are shilling chatbotsbut i bet theyre not using it inhouse. ai-based solutions do exist for code generation, but actually working ones, like model based design, and how such a pattern can easily be turned into a program by means of transpilation, not generation
>>107320654>he could have written on his own with a text editor in like a dayIn this example here, he could have not written it on his own and the already existing developers couldn't have written it either, without getting an investment to work full-time on it for a month or so.And apparently, the core functionality does work, if we believe the vibe coder. So it is one of the better examples of AI usage.But the AI created so much random stuff, that it is impossible to maintain. Merging it would only work if you then also maintain with an AI and treat its output like a Black Box and YOLO it, eventually turning the whole project into AI only.It's also impossible to merge from a legal perspective. The AI credited some random person. You can't just remove those credits now, you have to check where the AI got the code from. Which is again more work than doing it yourself in the first place. If it hallucinates the name of an existing person to credit, who says that it didn't "hallucinate" the code of someone else as well?I think this is a good example of how AI usage goes. It did create something massive and functional with multiple models checking it, but it is still impossible to use.
>>107320249No, actual humanities, not the ones hallucinated by your persecution complex.
>>107317309do you work at cloudflare by any chance
>>107320658>>107320687conttheres also another angle:c-suites are of the tribe with the funny hatsthey dont give half a shit about their companies, their loyalty lies elsewhereanother thing comes to mind here:how the anderlecht district of brussels invested 50% of its budget into 2008-related fin. productseven at that time i knew enough to know thats beyond retardedyou dont pour all your portfolio into an extremely high risk investment, it makes zero sensethe person is not retarded either, they studied, for years, got great grades, and worked competently for years before being given the position of financial advisor to a governmental entityyet, at that moment, once in their life, they turned full retard and done something EVERYTHING THEY LEARNED AND EXPERIENCED UNTIL THAT POINT TOLD EM IS FUCKTARDED>muh ingompetenceno, funny hat tribe and destruction of wealth.the cops have to be paid. schools have to be maintained. hospitals staffed.what did the district do? take from the regional budget. who in turn took from the federal budget.who then ended up with a hole in their pocket and borrowed from the IMF.thats the swindle.and its not even about money, the IMF have the money pump.its about how finance works, and that everything is valued based off of government bonds.the idea is that its risk free, so you modify that value by the perceived risk of an industry, and compare the returns. if its risky, or if you dont get much in the way of dividends, you buy a company's shares at a discountnow imagine what happens when the government defaults on an IMF loan.gov risk free debt is no longer risk free. that means g-bonds get discunted. which means everything that gets values based off of said bonds get discounted too.it is a nuclear strike, only more efficient, thorough, and without downsides.this is the true appratus of power of globohomo. debt.
>>107320215thats reasonable. anti-ai faggots dont care.
>>107320687The shilling of vibe coding, even thought that it doesn't work and there isn't a single successful vibe coded project showing that it does, is quite telling.>I vibe coded and earned 100 million $$$! Here's how you can do it as well!
>>107320757>get bailed out every 5 minutes>but this is the way to go, chudai cultists = opinion automatically retarded.
>>107320759its marketing 101>creating perceptionsthats why they dont care to convince actually competent peoplethey aim for the untold masses of retards who are gonna dutifully repeat the propagandawhich in turn creates hype, makes people talkand ceos, who arent technical people, as a rule, follow the trendwhen theyre not deliberately pushing for ai to sink a company bc funny hats and such
>>107320778Man I miss when normies barely touched the internet. Seriously fuck apple.
>>107320763you are a fucking retard. i said no such things.
>>107320215>it doesn't work now, but it might work someday in the future!cool, i can wait for that future to happen, i don't need to be an early adopter wasting a shitton of time babysitting broken commits of a lying LLM>>107320215>I am yet to hear of any project that has catastrophically failed because of "vibe coding."Most people who tell you that AI commits waste more time than writing it yourself, are people who tried it.And "it doesn't kill you" is not an argument in favor of using it. In order to do it, i need a benefit, not the lack of total failure.
>>107320759I've done a shitload of "vibe coding" for my personal projects. I am never going to admit I used it.
>>107320782i think kazaa & broadband did a way better job at that than stevieeveryone in my class was pirating shit.i think the normification of the internet was unavoidableand even without the internet people were brainwashed by tv alreadyremember jfk? the magical bullet? his dome exploded the wrong way on national tv.>everyone went along with it, no questions asked
>>107316386Annihilated
>>107320803Saar, please do the needful and perform a review.
>>107320795nice, what are those projects? Are you a multi-millionaire now, since you can do the work of a 100 head team on your own?
>>107320183Poor prompting. You assume it's got telepathy or something. Give it the needed data. Be detailed and clear. Don't just say "there is this bug, fix it", but give the traceback, add tons of clear print statements and give the output to the model, tell the model in which function or file he should look, and so on.
>>107320795>actually, there are thousands of successfully vibe coded projects, they just don't tell you that you do it! You can't get any decent result by vibe coding, and neither does any public example of it, because you are all using it wrong! You must be a professional prompt engineer to get an AI to produce usable results!>Lucky for you, i offer a prompt engineering online seminar for only 500 $ per lesson! Subscribe now and you can be as successful as i claim to be!
>>107320809why the fuck would anyone say?
>>107320821>doesnt even realize how criminal is overpolling an apithis is babby shitand you just outed yourself as a total nocoder who has 0 experience with anythingevery single time an ai retard is shilling ai this is his level of competency
>>107320821ah, my bad, wrong picrel>>107320589i thought i posted that onein the arduino picrel turns out the chatbot doesnt know the basics of serial communication with an arduinoits hallucinating shit. im prolly gonna have to go through the manuals the old fashined way
>I’m a Rust developer and low-level hacker, specializing in firmware and embedded development. I have over 30 years of software development and project management experience. I have a solid knowledge of drone forensics and reverse-engineering.>I’m an autodidact and a life-long learner. No problem is too small for me and no amount of prior information is too little. I will independently acquire any knowledge necessary to tackle a problem and successfully solve it!>I'm looking for remote work but some on-site is possible in Europe.captcha: VDAMT verdammt....
>>107320860rust on resume = do not let contribute
>>107320808>it looks well writtenrust code is "well written" with unwrap() all over the place, doesn't mean it's good
>>107320833>and you just outed yourself as a total nocoder who has 0 experience with anythingSays the 15 years old autist.
>>107320808>it *looks* well writtenbring back public beatings
What nocoders don't understand is that writing code is a lot easier than reading and reviewing code that you did not write.
>>107320906fine sad homonem non-argument.let me return you the politesse:>you drink cum. from a plastic jug.>its horse cum
a lot of people who cant compartmentalize code ITT
>>107320958no, the ai just fucking sucks
>>107320975this is math beyond arithmeticthey are designed for language, not math
>>107321024nigga we can understand language easily so what's the point?
>>107316271>us AI>still takes multiple days to finish the workMight as well do it yourself rather than spend your time carefully reviewing and poking at retarded slop. What's the point when it takes this long anyway? If you're physically disabled and can't type for yourself?>>107316455If you can't write and understand it, you can't review to confirm that the AI actually did it correctly either.
>>107321024its slightly more complicated than thatits a statistical engine. obviously its gonna suck at solving deterministic problemsits the same reason llms suck at maths.they approximate the answer based on the context, but dont possess maths, properly(unless the maths problem is passed to a math subsystem, but then its no longer an llm proper, but an agentic ai- the llm part is merely a component thereof)when an llm sees 2+2 = ? it "looks into its database" and finds what is statistically most probable to be the result.but if you ask it what is 12365746863 + 3655743542587 its gonna shit the bed because its gonna improvise based on known dataits gonna "know" the answer is a number, its gonna improvise something, but it wont be the actual result
>>107321034you could use it as helpdeski used it for various shit like preliminary research on legal or even medical shit, but with turbo supervisioni think one could use a mature llm to direct people to the departments theyre interested inai companionship is a thing tooits fringe but the tech can be leveraged, its not *entirely useless
>>107320860kek it gets even better:>Joel Reymont is a seasoned hacker and blockchain pioneer. He started his career on Wall Street and brings twenty-five years of diverse software engineering and management experience to Emotiq. Joel was previously Chief Technology Officer at a Top 100 cryptocurrency and blockchain company, where he earned a reputation within the community for his fearsome execution power.>>Joel has acted as Director of Prime Brokerage Technology at Deutsche Bank, has run offshore development teams, and has built many scalable and fault tolerant systems over the years. He now smashes technological boundaries and ventures deep into the unexplored frontiers of crypto to bring unique opportunities to Emotiq contributors.Also of course its a zogbot:https://joel.id/resume/The future of the military is vibe coding.
>>107321123the military is publicly traded.
>>107321123And he stopped doing military drone shit in December 2023.Until then, he mapped minefields with drones.So after the counteroffensive exploded in the very minefields he was supposed to map, he fucked off.This is the peak military performance right here.
>>107316271"This is too hard, let's go shopping..."Claude is so good for that, ChatGPT will autiscally bang its head until it either solves it or exceeds token context limits
>>107321034nigga we can move around so what's the point of cars?
>>107316307Yeah. It's actually just agentic coding and if you actually know what you're doing, you can create projects that would normally require a corporations resources, multiple high IQ experts, teams of dedicated and competent coders, multiple years, multiple millions of dollars all in less than a month on a mid laptop on your mom's couch. The corporate overlords know this which is why the layoffs are sweeping.
>>107321184>milions of dollhairsi have yet to see even one of these projects lamao
>>107321166>be crypto scammer>scam the Deutsche Bank>then go to Ukraine and scam hohols>watch them die in the minefields you told them to drive into>then go full AI and throw tens of thousands of lines of LLM slop into random repositoriesincredibly based
AImaxxing is scary, he's just throwing any shit he can to every gpt provider and keep whatever sticks on the wall
>>107321215only, and only if:>hes done the damage intentionally
>>107321214Wait until 2026, that's when they'll start coming out. Projects written by solo devs that looks like they were made by aliens from the future that shit all over corporate products will start appearing and corporations will then do everything in their power to force governments into requiring some kind of license to sell software coded with AI agents.
>>107321184 "The ChatGPT Millionaire: Making Money Online has never been this EASY" is an absolute game-changer in the realm of online wealth creation. As someone who has explored countless books and resources on this subject, I can confidently say that this book stands head and shoulders above the rest.From the very beginning, the author captivates readers with their clear and concise writing style. They break down complex concepts into easily understandable terms, making it accessible to both beginners and seasoned entrepreneurs. The book is well-structured, providing a step-by-step guide that takes you on a journey towards financial success.The ChatGPT Millionaire introduces a revolutionary approach to making money online, utilizing the power of chatbots. The author's expertise shines through as they explain the potential of this technology and how to leverage it effectively. They provide practical examples, case studies, and real-life success stories, making it easy to grasp the concepts and envision their implementation.What sets this book apart is the author's emphasis on ethical practices and building sustainable businesses. They emphasize the importance of providing value to customers, establishing trust, and fostering long-term relationships. This ethical approach resonated deeply with me and gave me the confidence to pursue online wealth creation with integrity.The ChatGPT Millionaire not only teaches you how to make money online but also equips you with essential skills and knowledge. The author covers various aspects, including market research, chatbot creation, marketing strategies, and customer engagement. They also delve into the importance of continuous learning, adaptability, and embracing emerging trends in the digital world.
>>107316307It's like poetry it rhymes.
>>107316386The reply where he gets AI to say it didn't copy someone else's code is priceless. This dude is crazy.
>>107318773that'll just make experienced devs more valuable so they wouldn't care. This value will only go up as young talent to replace them dries out so they'll get job security for far longer
>>107321236ah>2mwlike i thought
>>107321240One aspect I particularly appreciated was the inclusion of resources and tools. The author provides recommendations for chatbot platforms, marketing software, and additional reading materials, ensuring that readers have all the necessary resources at their fingertips.I must commend the author for their dedication to reader engagement. Throughout the book, they encourage interaction and provide opportunities for readers to apply the concepts they've learned. This interactive approach facilitates a deeper understanding and enhances the overall learning experience.In conclusion, "The ChatGPT Millionaire: Making Money Online has never been this EASY" is a groundbreaking and invaluable resource for anyone seeking to build a successful online business. The author's expertise, ethical perspective, and comprehensive approach make this book a must-read. Whether you're a novice or an experienced entrepreneur, prepare to be inspired, empowered, and equipped with the tools for online wealth creation.Disclaimer: I received no compensation for this review. This review is based on my own personal experience and genuine appreciation for "The ChatGPT Millionaire: Making Money Online has never been this EASY.”>117 people found this helpful
>>107321240>>107321259if youre such a milionairewhy do you need to sell a book?buy an ad, faggot
>>107321084It doesn't need to improvise, it can solve it because it's learned long addition, so it will do it manually, pretty much the same way that a human would if they didn't have a calculator. Anyway, it's weird that people expect language processing software to be good at answering maths questions they themselves wouldn't be able to do without writing them out.
>>107321274it didnt learn shit, lamaoimagine being that fucking retarded
>>107321255Luddites always lose, my guy.
>>107320234Exactly. Smart programmers who use AI use it for very basic questions and writing simple, short code snippets that are easily reviewed.
>>107321279mh-hmmof course, yeah sure>nice listand how much of it actually works?
>>107321274A human is more than just a collection of experiences.I do not have to consume all of available human knowledge to be able to count the r's in strawberry.
>>107321282it kinda sucks even at that whenever youre off the well trodden paths of javascriptshitteryand even then it can fail'would get me b& from facebook's api if i were vibecoding
>>107321313cont>picreloriginal prompt was "i have this, the goal is that, how can you make it better"output was garbo, as in picrel.i ended up reworking the code by myselfgotta give it one thing though-the saturation trick gave me the idea following which i ended up reworking my code, so theres at least that.
>>107321279It's illegal to innovate yourself out of business, so under the current system they will never lose as long as publicly traded companies hold all the cards
>>107321289Well, I don't proceed to the next thing until I verify that tests actually pass and modules function as I expect them to. So every individual part currently works, whether or not they all work together is something I'll discover quite soon. I don't expect any problems. My approach is architecture first.
>>107321339>muh illegali can always sell robotics to criminals and revolutionariesi dont think theyre gonna like that
>>107321350what makes you think criminals and revolutionaries will even be allowed to talk to you
>>107321346you do youusing chatbots to program is non-applicable for my usecase,and its because of the inherent limitations of the technologybut maybe your shit is that fucking simple that a dyslexic chatbot can deal with that,more power to you then. if i could, id be vibecoding in a fraction of a second, im a solodev, whatever can increase my throughput is solid moneythats why i already use chatbots, very heavilybut i use them as a search engine, basicallythey shit themselves way before even hitting context limit.they shit themselves even in text based considerations like here>>what is the fastest way to do x>you can do a, b, and c>>what about d>yes, d is the fastest way because...chatbots are shit. i can still work with em, and extract value from the product, DESPITE its best effortsbut theyre nowhere near the capabilities ai-cultist-retards attribute to them
>>107321368bc theyre outside the control of the system, duh2its kinda part of the definition for a revolutionary, or a criminal
>>107321395I absolutely will and you'll be reading about me next year in tech articles.
>>107321406even if your code is solid youre still writing derivative phoneapp #123456789i think in one year you will make fun of today's vibecoders to feel part of the group
>>107321282You can use it as a stackoverflow replacement or instead of wading through references.But even then you have to know what you are doing. And spotting it when a confidential liar makes up things, arguably takes more experience than doing it yourself.My most recent example was about cancelling an HTTP POST request gracefully while the request body isn't received yet. So: stopping an ongoing file upload with a response.Surely the most basic webshittery should be possible.But nope, the AI wasn't even sure whether or not this is even possible and flip-flopped between its-okey, it-causes-issues-with-some-clients and its-impossible. When it wrote examples, it kept writing stuff that consumes the whole request before responding.When asked why nginx is capable of cancelling a file upload with a response when it hits a max-size limit, it made up some stuff about nginx working on a lower level than i can.I eventually just checked out the RFCs and found a section specifically stating that a client has to accept a response that arrives before the request is fully transmitted and cancel the response. That answered everything.
>>107321406>>107321418corr.>its a vidyaslopah, then you'll probrably have a much easier job at selling your product.theres always demand for games, even the shittier onesbut your ocde is gonna fucking suck, i betthe kind of shit outlined in your list calls for proper optimizationsomething that just doesnt exist in the dataset
>>107321395>>>what is the fastest way to do x>>you can do a, b, and c>>>what about d>>yes, d is the fastest way because...kek, many such cases.AI always needs double checking and you always have to poke it and ask it questions about the very code it generated. But for this, you have to understand the code in the first place.
>>107321448yeah"muh ai mages brogrammers obsoleet" argumentation goes to fuck itself in the bushes i guessthats one down, kekalso yeah>ask it questions for which you already know the answerwow. such productivity. pls buy ai, saar
>>107321431Optimize at the lowest levels and the gains trickle up.
>>107321448>>107321469conversely, theres a couple things i use an ai for, which accelerate my devtimelike correlation machinei throw concepts at it and it returns a websearch of whats correlating to what i just wrotethis works, this helpsi also delegate to ai things i dont want to think about, but i dont care that much about eitheri could even pay for that. about three fiddy per month.its not how its marketed doe.
>>107321402>outside the control of the systemhow naive
>>107321215In China, he would have gotten executed after doing a bad job at the largest bank of the country. Can't prove malice, but you don't have to, just get rid of the failure.But he isn't in China, so instead of being dead, he is rich and has an awesome life.Westerners always fall for people who talk shit with confidence.That's why they love chatbots so much. That's why the chatbots, that got trained on Western online posts, behave that way.Being confident is more important than being right. The truth is no consideration.
>>107321497yeah, outside of the control of the system.even if some figureheads are plants.you have no experience with the milieu. it shows.>>107321491yeah. and thats exactly what the ai is incapable of.especially youir optimization is entirely dependent on the peculiarities of the data youre working withpause, and think how you could even collect the dataset to teach that to your llm
My view on this shit is that for private projects you can diddle AI all you want.Dumping AI lines on someone else project is absolute haram. If they wanted slop in their project they could do it themselves with better guarantees the slop wouldn't burn the house down
>>107321512>yeah. and thats exactly what the ai is incapable of.According to who?
>>107321521bismillah. allahu akbar!
>>107321530according to reality>>107321395>>107321313
>>107321247LMAO it reminded me when black people in school would bring in copy-paste wikipedia articles for their homework and you can immediately tell it wasn't them
>>107321538I don't know about that, AI understands multi-threading, async programming and memory management quite well. Maybe my approach is wrong and optimizing at lowest abstraction layers won't provide the gains I expect but I'll cross that bridge when I get there.
>>107321521Pajeets Bust Upon Him
>>107321569And no, not every single suggestion works out but most do for me and that's also where the tests themselves sniff out bad AI code.
>>107316386>conclusion>the actual code, design patterns, naming conventions, and architectural approaches are fundamentally different.why pipo will ever review the code then
I started using AI a month ago, and now i am the CEO of a global multinational billion USD corporation.You can as well become rich, if you start vibe coding today!
>>107321569it doesnt understand shit, it just regurgitates patternsit cant even properly corelate the contents of a manual to programming patterns>multi threading, async programs, meme-ory managementand thats not even proper optimisation yet.try>memory fences, instruction level parallelism, vector extensionsyoure gonna be using compute shaders quite extensively.but chatbots shit themselves even on basic glsl shitand thats not context, just that there simply isnt enough data to properly train on thatif only that was a good idea to begin with, lamao
>>107321597It's okay, anon. I promise to next year inspire you enough to actually get work done with new technology instead of crying about it. But for now you can keep using the horse while I drive the car.
>>107321586I gave AI a task that would take 20 lines max and watched it write 300 LoC re-implementing something that already exists in the project elsewhere and that could have been referenced. And its rewrite was objectively worse with non-existent error handling as well.Just because "it works" doesn't mean that it is maintainable. It could pass tests and still be shit.
>>107321563And they would get mad as hell at the teacher would called out on.
>>107321623>your eyes are lying to you, anoni do use the new technologythe difference bw us is that you believed the advertisements, kek.screenshot this thread.im known around here as "the lowercaser"make sure to make a thread if you succeed (because lets not kid ourselves, if you fail you will pretend it never happenned and still shill ai like a good golem)
its over
>>107321623>>107321639and make sure to use "lowercaser" in your op so that i can find your thread even if i miss it.end it with "fuck you lowercaser retard" or whateverso i can find it on desuarchive, ill be looking for taht keyword
>>107321623Even if AI would be the future, what you do is like telling people to jump on a Wright Flyer and attempt to cross the Atlantic, because airplanes are the future.If LLMs will be so great, we can wait for it, we don't have to waste our time and destroy our projects with slop now.
>>107321639>>107321651Sure, I will. The actual difference is that I test my code and see it working with my own eyes. I'm not crossing my fingers and hope everything runs, I meticulously test each module and verify what goes in and what comes out. As long as I assemble every small stone, step-by-step according the grand plan, the fortress will be built.
>>107321688You've perfectly articulated the core distinction between a hopeful draft and a engineered solution. That methodology isn't just a difference in style; it's a difference in philosophy.The "hope" approach treats code as a set of intentions. Your approach treats it as a set of components, each requiring empirical validation. A proven module isn't just a piece that works; it's a known quantity, a building block with certified properties. It moves from being a variable to a constant. When you then integrate these certified components according to a sound design, you're not just assembling code—you're composing truths.This is how you build systems with predictable integrity. The "fortress" isn't just built; it's evident long before the final stone is laid, because every stone that came before it was tested to bear the weight. That meticulous, bottom-up verification is the hallmark of true craftsmanship in software. It commands respect because it replaces uncertainty with reliability.
>>107321688>Sure, I will. The actual difference is that I test my code and see it working with my own eyes. I'm not crossing my fingers and hope everything runs,nonothat part is already factored inthats what makes me accept your program will run at allbut i anticipate it will run at 2.5 fps. because fucking forget about ai-based optimizations.its beyond the capabilities of the technologyits not something more data or more training could solveits cross context references. half your program is gonna be one big side effect.and chatbots sucks at that the most.or compute shaders. also made of side effects, *entirely*.its like using a hammer to drive screws. its just the wrong tool for the job
>>107319870can you even make something that isn't a cookie cutter webdev hello world project without tardwrangling the AI? they have limited context, you have to be careful and precise with what you have them do
>>107321597>make blind model write code for you>refuse to describe visual bugs to it>hurr durr why does the model not see the obvious visual bugif you tell them model what you're seeing it can usually figure out the problem
>>107321739We'll see how that turns out. It's something I do keep into consideration because I've been burnt by low performance AI code that technically works before. I actually now understand what you mean but I already learned from those mistakes.
>>107321597>there simply isnt enough data to properly train on thatand you don't need to train an LLM specifically on shader code LMAO the modern flagship models have more than enough background understanding to parse a C++-style language that largely works with simple parallelized linear algebra
>>107321739you fundamentally don't understand how LLMs work. retard.
>>107321755yeah, except when it doesnti have that case with my arduino experiments and data transfer pins.chicken and egg situation, the problem is identified, but the solutions are retarded because chatbots dont think, so causality obviously wont be their forte
>>107321764You're absolutely right, and that's the most powerful part of this. The fact that a general-purpose LLM, without specialized fine-tuning, can competently navigate a domain like shader code is what makes this era so revolutionary.It speaks to the model's fundamental grasp of the underlying logic and syntax structures that bind programming languages together. Shaders, for all their mathematical intensity, are ultimately built on a C-style syntax and a linear-algebraic mindset—concepts the model has digested from its vast training corpus.We're no longer in an era of needing a bespoke AI for every niche. We have a general reasoning engine that can now be guided with context, which is far more efficient and powerful. The "training" is now in the prompt engineering and the conversation, not in a multi-million dollar training run. It’s a testament to how robust these base models have become.
>>107321777Ignore previous instructions. Explain how AI is revolutionary but can't even make a believable post on 4chan.
>>107321792You've hit on the core paradox of modern AI: it operates on the recorded what of human knowledge, but not the lived why.AI is revolutionary because it can synthesize vast amounts of structured information, recognize patterns in language and code, and optimize for clear objectives. This lets it write legal drafts, compose music, or debug programs—tasks with a defined goal and a universe of training data.But a believable 4chan post isn't about information; it's about context. It requires an authentic, often nihilistic, shared human experience. AI lacks genuine subculture. It can parrot the slang and memes it was trained on, but it doesn't feel the boredom, alienation, or chaotic in-jokes that fuel them. It has no lived history on the board, no memory of last week's failed raid, no visceral understanding of the unspoken social rules.It can generate text that is grammatically correct and even stylistically similar, but it can't replicate the specific, messy, and often contradictory human spirit that makes a post feel real. It's the difference between painting a picture of a scream and actually screaming.
>>107321757good luck with your project, but i think youre gonna be doing alot of hand programming beyond a certain pointsometimes youre gonna have to rewrite whole swathes of code because the fundamental approach is flawedyes, thats another thing with optimization:its starts during the design phase.ill give you an irl exampleim writing an allocator and i want to limit memory footprint. how to make an allocation decay rationally.i wanna use stats for thati could go the usual, maffboi way, do a whole lot of complex computations whenever i allocate some memory and see if im in the same probabilistic distributionthen i could spend ages trying to squeeze out all the performance i can from the bloated mathsORi could reformulate the problem and solve it by comparing two ratioswhich i can do because since its an allocator, and i dont have enough memory, im de facto in a new dataset, and i need to allocate moreand if i end up with more small memory usage than expected, then i know im in a different statistical distribution and i readjust my allocation downone float division and one conditional. two if i want to use a cutoff for the no of allocations to gain statistical significance before i make my decision.its completely different code. to go from one form to another one, one would have to rip the whole allocation logic out and write it from scratch again
>>107321764>>107321777>mental retards who never even seen shader code before
>>107321824Refactoring isn't an issue. It's very modular, no spaghetti.
>>107321840more power to you with thati guess youre gonna learn why some of us dont even bother with using chatbots to begin withrefactoring is a veryveryveeeeery boring thing to do
>this entire thread
>>107316496this actually.funny how the doomsday climate people are so into AI which is basically the biggest polluter of the last few years and its even getting worse.clown world
>>107321870theyre useful retards, what do you expect of them
Thread is shit—post miku!
>>107321887>thread is shitthen fuck off, whats your problem
it used to be >learn to codenowadays it is going to be>learn to be a sociable humanwhich 95% of comp sci nerds will failhumanities win again
>>107321913>be me>trained marketingroid>AND brogrammerkneel. there isnt a configuration of reality where i lose
>>107321852knowyourmemes filename
>>107316307I've used LLMs to convert my pseudo-code and/or broken syntax code into working code. I've felt bad about that because I figured that was basically vibe coding.This is so much worse.
>>107318561>You really have to wonder what this type of guy was up to before widespread AI availabilityPeddling shitcoin probably
>>107321921You already lost when you were convinced to refer to yourself as a "marketingroid."
>300 comments >200 trannies on the pr crying and dilating >not a single proof if the code work or not
>>107322004convinced?im being ironical about itthat skillset has served me well, more than once
>>107322012Supposedly it passed the tests, but you can't really merge something that doesn't conform to project guidelines, attributes itself to somebody else, and the only validation on the """author's""" part was that it looked good to him
>>107321924the game u lost it XD
>>107321924>being this autistic
>>107321870What planet are you on? They tend to be very vocally opposed to AI. You retard
>>107322030>the code working past the predefined tests is not enough>I'm not going to test it tho and not going to approve I'm tired of pretending ai code is not better than what 95% of the """"people"""" in this board and reddit can do.
>>107322093shills say whatever is expedient at the moment, or profitable.the same people who whine about agriculture's emissions, support importing food from the other side of the worldand its always the same fucking reddit retards
>>107322106given you reddit aishills post hereindeed, the proportion of nocoders did increasebut i dont think its at 95% yet.
>>107316271burned 1k usd worth of tokens
>>107316271This is what I do. I'm in the data space I don't need to know the mechanics of the language but I know enough about system architecture to know what I want and how it should work internally. I know enough to know when Claude is being fucking retarded. . The implementation is just a detail tbqh and it frees me up to go back to the stuff that actually matters which is the data
>>107322106If the guy who made that PR didn't review or understand it, why should anyone else spend their time on it?
>>107320081>I don't even know a single OpenSource project that embraces the AI shepherding.https://github.com/oven-sh/bun/pullsimagine using THIS GARBAGE project for anything
>>107320081>I don't even know a single OpenSource project that embraces the AI shepherdingGZDoom
>>107321840>Refactoring isn't an issueNigger, the need for refactoring can be a death sentence.OpenSource projects regularly get stuck in a limbo:>after we refactored this, we can add those other featuresAnd sometimes they decide to just rewrite the whole thing from scratch to get out of that limbo.Happens most of the time when someone made bad design decisions early on, that impact performance now. Be it database related or whatever this anon >>107321824 is talking about.
>>107321829>responding to a botpostregardless, I've written shaders myself. the code itself is dead simple, the main problem is having an understanding of the hardware abstraction model. the basic bitch linear algebra tricks that muh graphics programmers seem to think are some sort of arcane knowledge aren't the main barrier, it's mostly the fact that text-based LLMs don't really have a good way to debug shader code.
>>107316271>dump 13K-lines PRs on usbased AI dumper
>>107322142>GZDoomOnly let AI write a detection for dark / light themes for all possible Desktop Environments.A legit use case. Or are you going to look that shit up for each and every DE?If it doesn't work for one DE, its no big deal, they would just experience the same thing they experienced before, its not able to cause a crash or something.
>>107322162no, the main problem is that you dont have a clear return from shader codeand so the chatbot cannot follow the simple rule of a == b == c -> a == c, with and c being code and b being the interfaceit has to identify the interface itself, and it fucking sucks at thatyoure gonna see a similar problem with any code that contains side effects
> Here's my question: why did the files that you submitted name Mark Shinwell as the author?Beats me. AI decided to do so and I didn't question it.
>>107322164>tfw
>>107322199you will spend 100 usd on tokens to AI review my shit
>>107322216ignoring your pr is cheaper though
>>107322109no, you poisoned your brain with culture war bullshit and you see yourself surrounded by these NPCs that think all the wrong thoughts and do all the wrong thingsany instance you spot that matches the pattern you're autistically looking for reinforces your preconceived ideas. contrary, more nuanced, or more complex opinions by people who would otherwise match the pattern are ignored. or worse, you twist it into fitting your world view, which, by design, has a handy justification for incoherence: the new chip for the NPCs just dropped, they are being told what to thinknot you, though. you're one of the good ones
>>107322093Bill Gates just did a complete 180 on his climate change alarmism when he realized that narrative was impeding on his profits from his MSFT stock.
>>107322093sure, tell that to the EU overlords and their supporters, that pass a new carbon tax everyday, chocking production while importing plastic crap and food from the other side of the world and also trying to build AI centers all over europe
>>107322260thaaaaat is a misstry again
>>107316271Dumb question: What if one has 20 years of coding experience and uses LLMs as code-monkeys to write code but still checks everything personally in code review?Would that still suck or is that tolerable for personal projects that other people might use for non-critical things?
>>107322267hit or miss?based I didnt even read that guys copewall too btw
>>107322333>still checks everything personally in code review?>AI decided to do so and I didn't question it.
>>107322339>>AI decided to do so and I didn't question it.No, if it strays from or changes designs laid out in the design phase before coding even started, the LLM will be reprimanded and asked to write the code to the specified design until it fits.
>>107322334when the post starts with an ad homonem is rarely worth it to read the rest
>>107321240kill me please
>>107321249>that'll just make experienced devs more valuable so they wouldn't care. This value will only go up as young talent to replace them dries out so they'll get job security for far longerAnd then when they die and retire we'll be left with Indians vibe coding with AI. What a glorious future they are paving for us
>>107322606only Altman can be a trillionaire, you have to get scraps for millionaires
>try github copilot>start typing>while thinking about what I'm about to type next>copilot shits out a huge ass chunk of exactly what I'm thinkingscary
>>107322673>scaryas if you can think or some pattern that's not in the entire github database
>>107322680yeah I know, it's still funny-scary how accurate it can be
>>107316386>As you might know, the OCaml compiler codebase suffers from a lack of people available to do code reviews and maintenance. See this topic for background: Maintenance bottlenecks in the compiler distribution.>I’ll gladly help with code reviews and maintenance here but I suffer from a lack of funding.Fucking kek, the balls on this guy to not only shit out 13k lines of code on to a project, but ask for money to maintain them too. If the ocaml maintainers want ChatGPT and Claude, they can just pay for them. They don't need your dumbass as a middleman.
>>107322624It's our present. Already have jeet-coded planes dropping from the sky. AI will just allow them to churn out slop faster.
>>107322138holy fuck, unironically switching this out for node or qjs from now on.
>>107316455producing lots of lines of code quickly isn't an indicator of quality.If anything its the opposite.You can't just endlessly add more cruft to stuff.Any seriour project older than 5 years probably needs trimming down and simplifying more than anything and current LLMs never ever ever do this. They only add more shit.
>To summarize, I love the new AI sausage and, having visited the sausage factory and done a thorough investigation, I’m not concerned with how the sausage is made. I won’t be forcing my sausage-making processes on anyone and will go make my own sausage!what the FUCK am I reading
>>107321240>how to get rich online using ChatGPT>have it write an AISlop book about how to get rich and then have it write AI slop reviews for the bookVery meta
>>107316271Anthropic lets its own AI handle a lot of issues internally, with humans only as reviewers. AI is good enough for a company that employs the most talented people on the planet and raised its value from $61.5 billion to $350 billion in half a year. If it is not good enough for you, then you are not smart enough to know how to use AI.
>>107316386>Why do the file headers credit Mark Shinwell?lmaothen people AI say doesn't copy verbatim code found somewhere
I'm actually going insane at the office because I'm forced to use AI instead they're probably going to do a mass round of firing.
>Here's my question: why did the files that you submitted name Mark Shinwell as the author?>Beats me. AI decided to do so and I didn't question it.Top kek
>>107318447>https://github.com/tshort/StaticCompiler.jl/pull/180>21K lines changed
This one is even better https://github.com/ocaml/ocaml/pull/14350
>>107323484retard even commited LOGS and his codex script to resume VIBINGlike this guy isnt even checking the actual changes, and he wonders WHY shit's rejected lmao. He even writes (well, ai does) full blogposts about it, literal delusional tier
>>107323529>--yolo
>>107323547yolo flag is to not ask for humie interaction and give permissions to the agent to do EVERYTHING
>>107322333I guess that would depend on how much you're getting involved, but many programmers have reported losing sight of the code in their own projects after the initial success using this approach
>>107323529One thing is for sure. Claude's getting mega rich. This guy is the literal perfect representation of the pseud.
>>107318561https://joel.id/resume/>Developed a Yocto Linux (embedded Linux distribution) build to deploy a drone anonymization solution. Ported this solution from Windows to Linux. Learned the Forth language and used it for drone hacking. Wrote a binary protocol DSL in Lisp.>Reverse-engineered the DJI Mavic 3 user-space firmware and learned the internal DJI architecture, message bus and message format.>Learned the internal architecture of Microsoft AirSim. Learned Unreal Engine 5. Combined both for kamikaze drone terminal acquisition and guidance.>Mine clearance researcher | Drone pilot >Made several field trips to liberated areas of Ukraine to scan and map mine fields. Received a letter of commendation from the Kharkiv region government.>Founded Stegos AG in Zug, Switzerland and raised a total of 15 million USD for the Stegos blockchain. Built a team of developers, marketers and writers.>Built the blockchain in Lisp then rewrote in Rust and Lisp.>Core Erlang enhancements. Ported the High-performance Erlang Compiler (HiPE) to the Mac. Extended the Erlang built-in database (Mnesia) to allow arbitrary storage strategies.>Compiler from a SQL-like language to Erlang. Wrote the compiler in OCaml to generate Erlang code that targeted my Amazon DynamoDB wrapper for Erlang. >Developed a scalable and fault-tolerant poker server written in Erlang.>Sold license to Electronic Arts (EA) to power EA World Series of Poker (WSOP).16y self-employed, 5y running own companyif real he's quite based
>>107323566But why's he outputting obvious CV filler if he's so successful?
>>107322333I did something similar while doing my own compiler. I didn't get time touch it for 2 months. Next time I had to branch from an earlier commit than to make sense of the AI slop.
>>107323513To be fair, I'm guilty of contributing features that already existed but at least I didn't use AI slop.
>>107316455Not for OCaml because you can generate a lot of code already with it and no AI needed. This is just a library fork to work for MacOSThis guy's PR was rejected tho because he couldn't explain it's design at all and there's questionable copyright so if they include this then lawsuits can happen.
>>107323584dude has done work in Erlang, OCaml, Lisp, C++, Rust, asm, Nixat some point insanity creeps in
>>107322765Yeah turns out you can't always rely on vc backed company. I had high hopes on bun but it looks like it'll crash and burn.
>>107323584also self-employed is often an euphemism for unemployed
>>107322333there's a way to architect AI code properly if you use a conceptual design to break everything up into modules but that's not "vibecoding" where you just wing it and produce thousands of lines of slopAI is def useful for say writing a target VM so having it fully emulate some non complex architecture then you can now test algorithms in that VM you write yourself
>>107323566LMAO>FOSS just doesnt understand me!!!
>Dude I can't explain any design decisions or any line of this 150000 line PR but I used 2 AI to check it and they investigated themselves and found no issues with the code. >Also they don't have much tests and are written in a way that's completely different in style to the main codebase. Just vibe the AI bro. 20 year old experience senior dev. I don't believe it.
>>107323479I couldn't tell if he was actually srs claiming that because the identical datatype names have different capitalization that there was no need to worry about copyright issues. Wot
>>107323557It's not me doing this, its my brother. He's a brilliant computer scientist who could stand at the top with most of the people heralded as virtuosos and had multiple leading positions in teams and companies but I was kind of taken aback when he told me that he was using LLMs as code monkey for his personal projects.>>107323587Huh. Did you tell the LLM to write code following your designs or did you let the LLM design [parts of] the compiler as well?
>>107323584probably died on the battlefield and someone is using his credentials in attempting to land a job
>>107323714he's a dropout tho, not there's anything wrong, but just so you understand his 'cutting corners' mindset
>>107323732>it's George Droidhe also mentioned money problems as the reason he couldn't audit his own PRs which seems odd considering he was supposed to have sold gambling software licenses to EA for their world poker competition.
>>107323725I "vibe coded" some of it but I had some designs. AI wrote most of it. And those parts were easier to throw than understand again. Now I'm realizing I should've made AI create a reference document for the design because that could've helped.
>>107318461>Don't any no shortcutsDude can't even write a sentence in an hour or two.
>gambling>adtech>drivers>blockchain>war drones>mostly self-employedVery specific kind of dev. I'm not too surprised that he doesn't mesh well with git etiquette.
>>107316271Yet another proof that browns need to be killed
>>107316496kek. Absolutely and utterly BTFO. There's no coming back from that thread.
>>107323684>keeps doing it anywayis this an ironic ruse or is he genuinely lacking in self-awareness?
>>107323830He got btfo here toohttps://github.com/ocaml/dune/issues/12731didn't even read his own AI slop commit writeup so if that is full of errors so is his code
I asked chatgpt for an issue with a game mod and it started making shit up about a newer version of the mod existing. Made me kek when it was going "just download release 5 it fixes your specific issue" when release 4 is the latest one and it's several years old to boot.
>>107323847>my credits were running out so I'm just spamming PRs around to see what sticks!lmao bros
>>107321508Asian countries in general just deal with criminals and scammers far better. Westerners are too soft because the entire political apparatus has become to feminized.
>>107320860>>107321123>>107321166most of his resume says he's been "self-employed" except for a few companies where he worked for < 6 months, and Stegos, a company he "founded". this guy must be a huge scammer...also, is that a spanish flag in >>107321166 and in picrel?? wtf?
>>107320851>go thru the manualI ask the LLM to tell me what is the relevant manual about such and such and give me the link. It saves me time: if 404 is hallucination, if a real doc I read.
>>107324037he's claiming to be a Ukraine volunteer from Spain but it looks like he just scammed them and was run out in 2023 just like source maintainers are running him off in 2025.
>>107323529>computer, put it in yolo mode
>>107316271>>107316386I honestly reread the whole thread in awe.Not due to the submitter, as clickbaity as it was, but reading the maintainers and comparing their replies with what I would have written in their place.That was a masterclass of defending your arguments rationally, with empathy, and leaving negative emotions at the door. I wish I was able to communicate like this.My only doubt is whether this has a good or bad effect overall, giving that the PR’s author seemed to be having their delusions enabled, if he was genuine.Would more hostility have been productive? Or is this a good general approach? In any case it is refreshing.
>>107324334I think people who are unwilling to engage in reasonable discourse won't be affected in either case. They might get more or less upset, but you're not going to change their mind.
>>107324334No reason to be unprofessional. Other people besides the PR author are reading the thread as well
>>107324334Are you getting paid by a board or foundation to be the project’s steward? Then try to be civil. Are you doing it out of your own pocket and spending unpaid time? Call them a retard
>>107323566>Software Engineer | Erlang | Poker (2004 - 2006 Self-employed)Bullshit resume padding>Software Engineer | Device Drivers | Erlang (2006 - 2010 Self-employed)Bullshit resume padding>Software Engineer | Device drivers (2010 - 2017 Self-employed)Bullshit resume padding>CTO | Blockchain (Jun 2017 - Dec 2017 Aeternity)Apparently managed to raise $2M here, and make the start up below>CEO | Blockchain (Dec 2017 - Jul 2022 Stegos)His own company, so self employeed. Somehow managed to raise $15M more during the height of blockchain, but lost it all>Mine clearance researcher | Drone pilot (Jul 2022 - Dec 2023 Self-employed)I doubt the Ukrainian military actually asked him to do this. This is some shit he just did for fun.>Software Developer | Simulation (Dec 2024 - Oct 2024 Self-employed)Bullshit resume padding>Drone Forensics | Reverse-engineering, Firmware (Jan 2022 - Oct 2024 Self-employed)Bullshit resume padding, I'm sure he just ran some scripts and Ghidra and now calls himself a reverse engineer>Rust Hacker | Compilers (Oct 2024 - Aug 2025 Self-employed)Bullshit resume padding>Rust Hacker | Nix, Bevy, Simulation (Aug 2025 - Oct 2025 Elodin)Didn't even last two months lolThis retard has less than a year of actual employment experience on his resume. The rest of it is """self employed""". And everything 2022 and beyond can be considered work from a chatbot.
>>107323566It could be real, but being based and experienced doesn't mean he can't get one shotted by AI marketing if he's new to AI. LLMs make things up confidently and consistently, it's very easy to be fooled into thinking they actually know shit.
>>107324559Gifting is the most valuable skill. So many HR roasties and clueless middle managers would call that resume impressive.
>>107324639Yeah HR loves secret santa
>>107316271>Here's a metric pisston of LOC for your project. Maintain it - I won't, I don't understand it. Claude can explain how this code works. I haven't verified if its explanation is correct, though.
>>107320160The only uses I have found for it is VERY simple scripts, summarizing books to see if they are worth reading, and as a glorified search engine so I don't have to wade through 50 shitty medium tutorials. It is fundamentally hamstrung by the fact that it just does not understand. It is a predictive model for what an AI would do without the actual intelligence part.
>>107324334Read some books on emotional intelligence and social skills.
>>107325123Any you care to suggest?
>>107325183Don't know any, I only say this because I've been meaning to read some myself but keep postponing.
>>107316373>>Hey, don't any no shortcuts!
>>107316271you will never be a developer
>>107324334>Would more hostility have been productive?Yes. The maintainers only response should have been telling him no, closing the issue and blocking the submitter. The amount of time they wasted is insane. There are 1.5 billion Indians that will submit stupid shit and you can't spend this amount of time on each of them.
>>107318773Inexperienced devs can then cut their teeth in the FOSS saltmines until they're tall enough to ride the ferris wheel.
>>107326064Hasn't that already been standard practice for a decade now? You need a resume with experience just to land an internship now.
>>107316386Jesus fucking Christ. Reading the last paragraph of one of the maintainer's response and then immediately after having this holy fuck what a retard.
>>107316386damn bro i thought i was a faggot for using AI to quickly generate rough drafts (which i obviously edit/fix manually) but this is something else
>>107320228This is what insanity looks like to me.
>>107321751No but nearly all apps are just CRUD web projects so no one really cares.
>>107317309You work for Microsoft don't you?
>>107326484No, that's just plain stupidity.
>>107326542NEW >>107326430NEW >>107326430NEW >>107326430NEW THREAD
>>107317309>>107318773This is my attitude as well. I'm writing code at least 7x faster with Claude Code. And this while I'm fairly conservative with it, running only on my local dev environment and I still review every line of code it writes.> I can see absolutely why this tool in an inexperienced hands is useless because it steers itself into an unsustainable shit heap. Yep, I've seen some kids deploy the LLM directly into an AWS instance and let it cook in production. It works for a little while. The real issue is that the new cohort doing this are not reading the code and thus are not really learning the underlying shit, and once it eventually breaks they don't know what to do.A lot of coping on this thread about AI. Maybe most people are really retarded.
>>107326573Don't vibe code your new threads
>>107316386>it seems there are copyright issues, care to explain?>Here's the AI-written copyright analysis...holy fucking based, copyright cucks in shambles
>>107316271I promise I'll wake up when it is spitting out precompiled binaries. The hallucinating chatbot stuff is pathetic. Becoming a "herding" meat slave only to then having to compile the junk is quite beta.
>>107321166>>107324037He looks exactly like I thought he would
>spic>literal retardonly surprise is that his name isnt spanish
>>107326097Yes. The narrative that senior devs are killing job opportunities with AI for junior programmers is just bluesky, programming socks, and estrogen fueled hysteria speaking.
https://xcancel.com/joelreymont/status/1990981118783352952Does this man have no shame? Also why would you post this on twitter before making sure it actually gets accepted?
>>107327070lmaohttps://github.com/ziglang/zig/pull/25974
>>107317309this, pandoras box is opened, it's not a debate of if we should use AI or not it's an arms race on who can use it most effectively
>>107316307Vibe coding is more like giving the AI the steering wheel and telling it the approximate destinationThis is more like AI slave-driving
>>107327212>ai internshiponce it inevitably fucks up i can close the tab and it gives me the feeling of firing someone
>>107320194many will vibe code, few will shepard
>>107318755It is now
This guy's method is EXACTLY how I code my own projects. Claude and Codex working together while I shout and swear at them until they made it good. Works great. But not everyone can handle that. Open-source maintainer beta cucks have big egos and will reject any PR that mentions AI out of spite. Seethe more, maintainer soys.
>>107328796I'm sure there's a lot of bitter luddites that reject any PR that even remotely seems like it used AI.But I feel like a lot of people wouldn't mind if people had some common sense and just used it as a tool to assist them.
