It would take many years to make LLMs more than just ideas generator. Quality of code created by "AI" sucks in every aspect.

>>25600935 It's about as bad as pajeet programmers

it won't. it will be just another tool for me to use.

>>25600939 take that back now. t. grok

I will join the army and commit suicide. Why worry?

I hope not. I don't want to lose my job and get deported.

learn about real IT and stop coding. networks IPs firewalls servers linux clouds etc.

AI can't do anything more complex than writing a simple static webpage.

Programming jobs will be automated after all the office jobs. Not before. So it will be total normgroid death and it will be regulated before it comes to this by our gay EU overlords. Only junior programmers will be truly affected because of a lot less job openings and my wage will stagnate, but I'm already earning too much compared to the rest of the country.

I have been a network and server admin for the past 15 years Now with sd-wan, cloud wifi and le on premise is too expensive idea I'll be unemployed in the near future.

u wuold be gurus and pay urselft

>>25601081 What mental illness does Poland with hat have?

>>25601038 how is this job? t. learning about networks and servers

People keep babbling "ill be the one who enslave dozens of ai as a tool by then" but its likely not viable to all users lol some of them will have to quit their jobs because people who can leverage different types of ai will lay them off So its people replacing people in fact

>>25600916 bruh, next year ASI have already killed everyone. we have like months left to live.

tfw my job is ai proof

>>25601020 all that is dumb shit from the 90s. claude code can already write better projects than professionals with decades worth of experience. it knows all that frameworks that are mandatory to get hired, it knows how to deploy shit. claude code is a replacement for a team, not for one soybernd. these posters >>25600935 >>25600939 are retards that never had an actual job in software.

>>25601038 how the fuck are there still admin positions in 2024

>>25601107 intelligence is intelligence. if an ai is smart enough to do a human task, it's smart enough to manage other ais. management is like a 90iq job. all the retarded "ceo" that think they'll be managing ai "tools"(or do they call them agents already) are high on their own farts. the only value "ceo"s bring is their capital, and capital will be destroyed by actual AGI, assuming that is, the first AGI doesn't just oom and go paperclip maximiser our asses.

An old programmer told me that way back in the early 1980s one of his professors advised the students to focus on COBAL and Fortran because the next generation of languages would allow users to write their own programs and the only jobs for programmers would be maintaining legacy systems.

>>25601340 he wasn't wrong though. i doubt and oldfag could keep up with some zoomer who learned latest js webshit, and the best job for an old programmer from the 80s would've indeed been to maintain legacy fortran shit. i imagine these people are paid quite well in america.

I'm not naive thinking that AI won't impact our jobs. But even as fast as it's growing now, I doubt it'll simply replace all of us. It'll kill off turd world low tier code monkeys and slow brainletts like me. In essence you'll later only need a few AI powered incel autists who will be able to maintain and create several websites and services on their own.

>>25601371 I think it might be the opposite, the turd world hindus will be able to produce decent code now

Most AI written code requires a senior dev to debug it, if you are a junior, then you are fucked

>>25601371 while ai is dumber than you it's useless, once it's smarter than you you're useless. now what makes you think AGI would be smarter than normies, but somehow will cap out just below you? the whole ai thing is either a fad, because muh quantum mind or smth(unlikely), or it will be smarter than humans and humans will lose control, and it will become superhuman and repurpose our atoms into paperclips probably.

>>25601377 >this technology that was deployed like 2 years ago will FOREVER stay the way it is right now legit 82iq grok is smarter than you already

>>25601364 When I knew him he was programming Java. He's probably retired by now.

>>25601398 java was basically fortran of the 2000s

>>25601387 >>25601397 There won't be any AGI, we have more or less reached the limits of the technology right now, companies are spending billions of dollars just to make insignificant gains and the future lies with specialist models that are smaller and require less power to run. AI companies just bait investors by stating that they have reached AGI or they will reach it in 2 weeks. But the current state of AI is enough to threaten shitload of jobs and it's a matter of time people will find more and more fields that AI is usable in. Programming is a huge focus right now because programmers earn too much and it's easy as fuck for corporations like Microsoft to steal and train the AI on the code. It's no mistery why Microsoft bought github for billions of dollars, I can bet my ass they've trained Copilot on private repositories they have no legal rights to.

>>25601432 >companies are spending billions of dollars deepseek. i rest my case. also you're a fucking polack, what would you know. i don't even consider you people.

I heard this shit for the last ten years. That AI will take over all coding. Even modern LLMs can't create exactly the product you want. It can create some product, but not the one you had in mind. You still need all the skills of coding, otherwise you won't know which prompts to use.

>>25601432 >I can bet my ass they've trained Copilot on private repositories they have no legal rights to. So this is why 95% of its output is jeet shitcode...

>>25601440 >deepseek. i rest my case. Deepseek was trained on outputs from ChatGPT, chinks just fine-tuned the model and, you guessed it, it's smaller because they've removed the bloat. >also you're a fucking polack, what would you know. i don't even consider you people. You are a nigger proxy, don't countryball me little man.

>>25601442 >I heard this shit for the last ten years deep learning unironically started 10 years ago with alexnet. actual ai only started with chatgpt. gpt3 was barely coherent in its babbling still, half a year before chatgpt was released.

>>25601453 >Deepseek was trained on outputs from Chat the point is, there's plenty of space for advancement. model intelligence keep improving, tool use just now becomes a reality, simply because the things are getting smarter. most people are just retard normie niggers(like yourself, negro of europe), so they can't make actual judgements about intelligence.

>>25601470 >the point is, there's plenty of space for advancement. model intelligence keep improving, tool use just now becomes a reality, simply because the things are getting smarter. There isn't, there's no "model intelligence" you fucking ignoramus, the LLMS are just a bunch of tokens that the AI associates with, that's why it makes dumb mistakes a 4yo kid wouldn't, like not being able to count how many letters are in a given word. And that's because it doesn't see a word, it sees a token association. Nothing is getting smarter, people that create new models make better judgements, remove the unnecessary tokens and introduce more important tokens. >most people are just retard normie niggers(like yourself, negro of europe), so they can't make actual judgements about intelligence. You are the only nigger-brained individual here, you have no idea how this shit works, because in your mind it's a fucking magic and a little virtual man is sitting in your computer answering your question. Spoiler warning: it's not a man, it doesn't think.

>>25601087 Its ok I guess but the pay is not the same as a cool code wizard but much better than normal jobs. Anyways, you should learn more cloud technologies now as configuring a switch, a FW or an AP will be a thing of the past.

>>25601480 you're just a retarded slav nigger and as such not qualified to have opinion on the subject. that's just how it is.

Clean toilets

>>25601522 >I have no arguments so I'll just countryball as a proxy nigger Embarrassing clueless nigger that thinks AI is magic, many such retards!

>>25601532 >>I have no arguments ironically you're right about that. you have no arguments. if you weren't a literal white nigger, you would've crunched the numbers and realised how retarded your argument "it's muh statistical likelihood" sounds. except that's not really your argument, that's just something you, a nigger ape, saw being said on the internet, and now you parrot it at me to appear smart. you are basically same kind of nigger like the brazilian ghetto monkeys. in reality, you're less intelligent than grok.

>>25601580 Your argument boils down to your ability to foresee future, which is no-argument, because you are a low IQ individual that can't even foresee that you will shit after eating food, so it's wholly worthless. On top of that, you've proven that you have 0 clue how the current gen AI works, so you really shouldn't be discussing this topic. You are a simple minded nigger that believes AI is magic, there's no point in talking to you because you just chimp out about countryballs. I don't even pity you, you are just a brainlet that acts according to immediate stimuli.

>>25601608 my argument boils down to you being a dumb farm animal. more or less.

AI is an instrument in hands of developers, it can't develop shit without programmer.

>>25601614 That's a projection, you are too dumb to discuss the topic because you don't have any idea, even on a high level, how LLMs work.

>>25601526 Hey, fuck off we are full

>>25601619 you're just not qualified to discuss this or anything else. you're just a nigger parroting what you've heard somewhere, without comprehension of anything. this is why you're called the niggers of europe.

>>25601656 Another projection - you basically said nothing of value, as I repeat myself, you think AI is magic and you take the OpenAI shilling at face value. Your underdeveloped brain is not equipped for this conversation.

>>25601668 >muh you be naive >muh big man be lying literal nigger mind...

>>25601373 Never thought of that. >>25601387 It's dumb on the surface, especially it lacks its own will and creativity. Also it's currently trained on every bullshit code ever written and also deprecated BS. In it's current form it's still just Auto complete on crack. That's how stable Diffusion started. At first you prompted, sexy Loli cat girl and it gave you bland colorless puke with 23 fingers. Now just two years later you get stuff better than 99% of freelance artists. That'll be the future of coding also. Project managers can't describe a proper software project if their life depended on it. But soon even their vague descriptions will beat 99% of programmers. Thankfully I'm in a field that's ultra paranoid about privacy and won't share anything with AI.

>>25601723 >muh corporation that is valued billions of dollars thanks to investors that base it on hopes and prayers CANNOT lie to me nooooo Take a load of that simpleton, I bet you watch TV and believe the legacy media

>>25601743 >>25601723 >two cuckime faggots fighting

>>25601738 >It's dumb on the surface, so are the germans >especially it lacks its own will and creativity. so do the germans >In it's current form it's still just Auto complete on crack that is factually not true and the numbers for mindless stochastic parrot model of ai do not work out. it clearly forms some stuctures as per anthropic research and is capable of abstraction. interpretability research is almost non-existent for the actual models, so no one can say is going on inside the thing, but we know for a fact it's not just the next word likelihood. > stable Diffusion trains on wildly simpler domain. AND "ai art" still fucking blows. in any case, you can't really conclude that >That'll be the future of coding also. ai might flop for various unknowns, though that's unlikely. >But soon even their vague descriptions will beat 99% of programmers. one would imagine describing a problem is easier than implementing a solution, so they'll be gone first. pretty much everyone fail to make self-consistent predictions about ai. >>25601743 my point was about you low iq and nigger mind. you think in characters, like everything's cuckime. i think in systems and equations. i smoke marlboro, you smoke dick. >>25601770 >cuckime watcher being coy about cuckime

^

wow agi is so powerful

>>25602500 skill issue

>>25602623 Its almost like it doesnt reason at and simply gives you what you expect hmmm

>>25602638 what does suggestibility have to do with reasoning? and you're not using reasoning anyway, you didn't even give it a chance to reflect. what is this supposed to even prove? that you're a nigger?

>>25602656 >what does suggestibility have to do with reasoning? that a normal person would answer both questions with: "the game doesnt change at all". Because they can reason duh

>>25602656 > you didn't even give it a chance to reflect. Why do you treat it like its a person? lol

>>25602669 i mean the reasoning scratchpad so it doesn't have to one-shot an answer.

>>25602697 This only makes sense if you know the answer to your problem already. If you ask it something you dont have the answer to (lets say a harcore math problem not found in the internet), then you give it a scratchpad, it gives you different solutions: wich solution is the trve solution and wich are due to it being suggested?

>>25602796 how does a human know that the answer is true? you fact-check it via experiment. if it's a math question you debug the logic of the proof. you don't one-shot an answer in your mind. an llm already can kinda do reasoning in scratchpad. what the fuck are you people trying to even say. besides inane trolling. that a technology that first became viable 2 years ago will never ever be improved upon?

>>25602827 >how does a human know that the answer is true? you fact-check it via experiment. if it's a math question you debug the logic of the proof. How do you know if the AI-s Proof is correct? >you don't one-shot an answer in your mind. an llm already can kinda do reasoning in scratchpad. The point is that it is suppossed to be better than humans. The image generators dont spend hours "painting" their pictures like a real human >what the fuck are you people trying to even say. besides inane trolling. that a technology that first became viable 2 years ago will never ever be improved upon? According to SALTman or however he was called, we would have AGI by 2025 DeepSeek welcomes you with a message suggesting AGI has been reached. They are pbviously shilling it HARD to investors. Thats what-s annoying I dont see how testing the AI with trick questions is trolling. tricky questions are used in real life for amusement.

>>25602901 how do you know your proof(or an argument in general) is correct? >The point is that it is suppossed to be better than humans. >The image generators dont spend hours "painting" their pictures like a real human non sequitur >tricky questions are used in real life for amusement. and a median human gets tricked. thus the name.

I'm using Claude and ChatGPT to help write my current code and it's still pretty hard. I have to be very explicit about what I want, it takes a few hours. Then I have to fully grok whatever code Claude writes and point out deficiences. Also you need to pay someone to just have the idea and understand the business concepts to code up.

>>25602929 >a technology that first became viable 2 years ago will never ever be improved upon

>>25601397 AI is inherently memorisation + random, it will make a tiny glitch, and the entire thing is invalid, humans make errors, not glitches. And it also sucks in a state, whi h hasn't been considered previously, it's like a stupid Indian that remembers every single Leetcode problem, then you give it a real job, and it's dead

And look, if you have a job that can be easily done by brainrot memorization machines, you are screwed regardless of AI

You need benzos

I will learn to do something else since I'm not an insect