>it only took two prompts to build a fully functional android appsoftware engineering is over. what's the next "learn to code"?
>>108343296Learn to think. Anyone can prompt an AI, but only a skilled thinkiner will come up with ideas, and only a skilled IT chad will prompt it via best practices and security
>>108343296Soon anyone anywhere that develops for Android will need to have their government Id in a Google database even if they never put anything up on the App Store.Look it up
>>108343357Apple already does that and has done it for years. Nothing new here
>>108343317Cope. Indians and Chinks will soon copy and steal whatever you vibecode.
>>108343296Knowing how to program is still indispensable. Not knowing how the code it spits out works and how to debug it when something inevitably goes wrong will only get you so far
Making apps hasn't been a valuable skill in over a decade
>>108343410just have the ai debug.humans make mistakes. humans debug. no different than ai.
>>108343296Writing code was never the bottleneck. We even had a term for people who's only job is to write code without any understanding "code monkey". So no it's not over. Enjoy your sloppy app tho.
>>108343357And there will be 3 billions ways to bypass it.
>>108343362And now there’s nowhere to run
>>108343724And then they’ll check every time you use the App Store or whatever
>>108343317That's like telling people to learn to run faster after cars were invented.
>>108343296>fully functionalnobody tell him
>>108343783not sure if you've actually tried using claude but the latest models have no problem one shotting almost anything simple-intermediate.
>>108343317security?
>>108343296>+277k additions, unmaintainable garbage that could be done in 10ki refuse to ever use any ai slop in my projects. rubber ducking only
>>108343296learn to suck dick
>>108343296>using code you cannot understand in production
>>108343803>the latest models have no problem one shotting almost anything simple-intermediate.Not my problem.Only really an issue for boilerplate code-monkey jeets.
>>108343296>277klocwhat the fuck did you write, an entire operating system with driver support for hardware dating back to 1977?
Still cannot code a Go board game app properly.LLMs cannot create a strong engine at 9x9, nor do they know the concept of resolving dead groups and counting territory.
>>108343296learn to vibe orchestrate
>>108343766Then i won't use it.
>what's the next "learn to code"?3D modelling. AI can't do it properly yet, but will be able to down the line.
>>108343296What does it do?
>>108343317>I am a senior thinkineer
>>108344891You are probably right for game assets and so on, but AI can already do CAD fairly well.
>>108344891I've recently saw on LinkedIn that at my former workplace they've start selling a vibe-CAD program. They already sold the service to a company which managed reduce their engineering team to a fraction.
>buildThat's a weird way of spelling "copy another project".
>>108343296uh, welcome to vibecoding? now use it and find the bugs and decide on some improvements to your code. come back and tell us how the AI does at bugfixing and adding all those new features.
>>108343296I expect you to be a trillionaire by the end of the month. Surely, the ai generated apps are usable right?
>>108343549>hey i need an app that does this>well then according to your requirements you need $25000, two coders, 3 months, 4 different types of contracts, bills to pay, taxes>Writing code was never the bottlenecklmaoing at this cope
>>108343549retard
>>108346329>hey i need an app that does thisHave you tried github?><incompetent autism noises>
>>108343775Yes? Look how valuable being a fitfag is, socially and economically. Modern life turned people into gross little goblins, the ones that choose to run and look good get infinitely more opportunities. Fuck, look at tech. All the successful white people are highly attractive physically. You get further in tech now by charismamaxxing and looksmaxxing.
>>108343296how's that security anon?
>>108346356>ignores all the other looksmaxxers who end up broke and in debtI can feel the IQ points that Gen Z is missing.
>>108346365Fat fuck dont @ me ever again
>>108346370Swiping doesn't work here, zoomtard.
>>108343296On average, at least in my case, every 200 lines of agentic code generated with Claude Opus 4.6 costs about $3, and the whole project is around 15k lines. I’m assuming that in such a large project, it’s closer to $10 per 200 lines. So that’s about $10-15k down the drain, without even knowing what’s going on inside. Investigating and reading all of it would take ages.
>>108346356>all the successful white people are highly physically attractive
>>108346400lizards aren't white.
>>108346390how? i use it daily and it's hard to hit limit unless i run them in parallel.
>>108346390lol? are your figures in indian rupees? this is nowhere near accurate cost of tokens.
>>108343462A "mistake" actually isn't really the same with AI, as what's more often the case is that it's correct about the wrong thing. If you actually know what you're looking at you can solve a problem much quicker before ending up with an AI having you chase a wrong solution without you knowing it because it has the wrong premise and you're just following it blindly. Someone who has the knowledge to do something without AI will get it done more quickly and efficiently with AI, than someone with a lack of the same knowledge, it will always make a difference
>>108347093nta, but we use the API in my company and I'm afraid to run the /cost command. Opus wastes a huge amount, especially with high-effort reasoning. If it weren't for caching, I would easily be spending around $10 per prompt.
>>108346329>>108346344unemployed
ok but can the AI fix it when it breaks?
>>108347674yes. developers are now qa who gives feedback to claude.
>>108348011i use reddit ai instead
>>108347674no. when you try to ask it to fix bugs it starts writing code that doesn't compile.
>>108343462The fundamental difference is that LLMs are not AI. Any actual AI researcher knows this. An LLM is a tiny piece of what a theoretical true AI system would look like. We've been in an AI winter all along because we're neglecting true AI research while we masturbate to LANGUAGE MODELS. Yes, they are great. Now we need to take a step back and continue the long path to simulating actual intelligence on silicon.
>>108348543oh look it's the guy that doesn't realize that language changes with usage and every single large technological corporation and every single educational insutiton on the planet refer to LLMs under the AI umbrella, which means LLMs are AI regardless of how much you cry/
>>108343296>300k loc for hello worldtech debt is over
>>108343357>what is adb installbut yea, it will filter normalfags
>>108349475>tech debt is overHeh, this is more tech chapter 11 or maybe tech chapter 7
>>108348543>The fundamental difference is that LLMs are not AI.They are, but they aren't AGI. They're part of what's needed, sort of, if they weren't so utterly shit to train.We need something to keep things consistent, we need enormously better training and attention, we need better handling of noise. And we need them to be faster and use a lot less energy (so we can use them in more ways).
Literally no good software has been released since the AI craze began.
>>108349534and before it?
>>108343296>a fully functional android appmay I see it?
>>108349660No
>>108343296Why are you so obsessed with fooling people into believing this? Will that make your garbage LLMs more capable, if you fool a couple of idiots here?What is your endgame, you retarded assclown?
>>108349765I think it's just the spite of the incompetent.
>>108343296Learn to suck willys
>>108343296Is that "functional" android app in this room?
someone told me to pay so that I can use sonnet extended, but sonnet extended is free.they said opus drains your account too fast.
>>108343296>what's the next "learn to code"?autistic knowlege of drone flight controllers and telemetry
i embraced the slop. i just sub'd to claude. i feel like such a simp.>>108351227so that we can blow up the datacenters?
>>108349977
>>108343296SAAR AI BEST! AI BEST SAAR YES! ISRAEL FULL SAPPORT PLEASE NO SEND ME BACK INDIA!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
>>108346329People are in the heavy cope at the moment>Hehe, just two more weeks amiright?>No, right now. It's good enough right now.>H-haha, actually you mean "two more weeks" it's always "two more weeks away" with you LOLAnd repeat. This despite the fact that most of them are already using it and just don't think about the implications.
>>108351187It's so you can use claude code. If you want to write large programs or work on a code base that spans multiple files it's pretty much necessary, the VM the browser version gets sucks.
>>108351773>It's good enough right now.>>108346272
>>108351807Why would I care if it copies form another project? I don't give a shit about "artistic integrity" with code, I'm here to make money
>>108351829You won't make money because everyone is using the superior original, whichbmust have been popular enough fir the AI to train on. And you could just copy the original and tweak the interface for it to look yours without having go pay 100$/month for it which is heavily subsidized anyway.
>>108351958Ah yes, if I just hire a team of SEs to do it instead and pay 3000x more to get the same result that will definitely be a better deal.How are you still not understanding?The continued cope is only hurting yourself, you know.
>>108343296git clone fully-functional-android-appOr, in Python:import fully-functional-android-appThat's one line bro, AI is cooked.
git clone fully-functional-android-app
import fully-functional-android-app
When I say AI you say BUBBLEAI
what are you going to do when AI companies need to actually make a profit, and the enshittification and rent-seeking kick in hard?
>>108351985Enjoy getting your pants sued off of you then.
>>108352032AI code has been a thing for 3 years, and agentic coding has been a thing for almost 6 months. So where's the lawsuits? Why would they start now?Wait, let me guess:>two weeks
>>108352029Government is subsidizing the fuck out of all of this, they'll never need money. DeepSeek has the entire budget of China to spend. Anthropic decided to morally grandstand and refuse that military check so they will probably go bankrupt or be bought out soon. Don't get attached to Claude.
>>108352044https://copyrightalliance.org/ai-copyright-lawsuit-developments-2025/
>>108352045you think china is going to be subsidising you and millions of westoids and a billion jeets in 20 years?
>>108352092lmao, this is people suing AI companies over training data. And the only judgments have been that they owe a pittance for using it. So again, why would I be sued?
>>108352123I'll bet that those of us who swear fealty to the great People's Republic of China will get free DeepSeek V9 access while the rest of you stand in line to rent public terminals to ERP with Llama 8b.
>>108352125Because you're using copyrighted material to make money off of it. It's that simple.
>>108352143So in other words,>lawsuits in two more weeks!Got it, thanks!
>>108351773It’s not good enough even if you take senior software dev salary worth of tokens and give them to said senior software dev though. Claude 4.5 fails rather simple and repetitive taks and I give hime fairly good instructions in a separate file. And then agent is OOT and the next agent you summon doesn’t understand shit
>>108344891this clueless guy lmao
>>108352720Don't bother. These people have a self-admitted, pathological hatred towards their betters. They cannot be reached by arguments, only euthanized before their hatred consumes them completely.
>>108343296>>it only took two prompts to build a fully functional android appTo copy one from existing repos, you mean.
>>108353252if you encounter a problem during software development, there is a very good chance that someone has already encountered it and has put the solution out in the web. even if the models are just fuzzily pasting solution from the internet it can solve 99% software engineering; when was the last time you encountered a truly unique problem that you couldn't find solution from internet?
>>108353495>sprintf locking despite there being no global resource it would've needed>SIMD hex/dec formatting and printing>zero-copy hugepage file mappings
>>108343296>low-effort bait
>>108343803>almost anything simple-intermediate.So nothing actually worth doing, got it
>>108353999i don't see the models plateauing. in fact we had a huge progress last november and i expect a few more big jumps since the ai companies can now capture the data from real world agentic use and use it for training.ironically software engineers themselves are helping to train a tool that will replace them.
>>108354338>i don't see the models plateauingDid you hit the methanol again?
>>108354379>Jul 2024get your outdated shit out of here, it's starting to smell just like you
>>108354422You're free to post something newer.
>>108354466All frontier models has already passed 90.
>>108354484Unlike you I haven't hit the methanol, but I don't see it.
>>108354338>i don't see the models plateauingThey already have retard. I feel sorry for you.
>>108354338>ironically software engineers themselves are helping to train a tool that will replace them.You could feed it an infinite amount of training data and it would still have all the same problems because it's a next-token predictor. I'm sorry that you're really jealous and insecure that other people know how to work with computers and you don't, but you aren't going to be able to live out your fantasies of being an epic hacker man with a glorified markov chain.
