All LLMs could vanish tomorrow and the world would still be pretty much the same.You can't say the same about any actual revolutionary technology.
>>109470635What would change if you disappeared?
>>109470698No one's giving me trillions of dollars.
>>109470635A few of my co-workers would melt down if they had to go back to work.
>>109470635Intelligence is not a "thing"stop being an asshole
>>109470834Yeah, I know. So I'm a senior software dev, 20 years of experience. I would hate it to have to code without LLMs for work. Personal projects or own business no problem, but solving someone else's bullshit and especially trying to understand someone else's retarded design and code... eww, I'd get really depressed, but maybe that's age showing.
>>109471264>especially trying to understand someone else's retarded design and code..so the very thing you have to constantly do when you let ai write your code?
>>109471301Actually no, that's what AI helps the most with. I feed it the relevant code and it's pretty good at giving me the high level picture of it and answering my specific questions about the codebase. Then it's much easier to write the code myself or guide AI. If I had to name one thing AI is excellent at its this: understand the shit someone else wrote with or without AI, making sense of it and guiding you in your work. Without this, it's easily 10x time and effort and if not more, I hate reading someone else's code and I hate most of the design ideas that are currently popular.
>>109471332You may ask, how dk I know it's good at this? Because I fed it my own retarded and sometimes unconventional code and projects and it was pretty good at explaining it back and guessing my intent and ideas behind it.
>>109470698>"n-no u"Fpwp
>>109471301>so the very thing you have to constantly do when you let ai write your code?Where the fuck did people work where all of this good, well thought-out code was produced?I would wager AI can produce better code, architecture and technological decisions than every single one of my coworkers for about 6 months now, and better than our Indian offshore devs for about 1 year now.We have boomers at our corp that say the exact same thing then they hand me some handcrafted piece of shit Java 8 spaghetti code that doesn't fulfil half the specs but at least they made sure they personally reviewed ever for loop to ensure they are 0.000000001ms faster for all of our 7 users
> All LLMs could vanish tomorrow and the world would still be pretty much the same.This would be true perhaps a year ago. But today LLMs are massively used by programmers pretty much everywhere. Also by other professions that deal with text.
>>109471264Retire and let people actually interested in working fill your spot, then. You're close to layoff anyways.
>>109470635Yeah that's why I've been saying banning AI wouldn't change anything.Except megacorps wouldn't be able to scam the whole world with their bullshit and they couldn't cut down on expenses with software devs
>>109471411AI is good at writing mediocre code and barely passing architecture. It literally cannot imagine anything better than whatever it was trained on (and it averages out to bad or mediocre code code) unless you tell it how to design it. If you tell it how you want it to be or discuss the options, then it can maybe write an ok code, but it won't be great code, because of the stupid data it was trained on.
>>109471431No reason they couldn't pick it back up the way they did a year ago. They'd just be rusty for a bit and probably need to hire back to pre-AI employee counts. Unless they let the AI spaghetti code progress itself to the point of being incomprehensible, but they shouldn't be letting that happen in the first place.
At my job a lot of the infrastructure is cooked and almost impossible to debug as it is. I have no doubt GPT-5.6 would improve it significantly in an afternoon. But we are very restricted on usage.
>>109470698Nothing. However I don't cost $x trillion or claim to be world-changing technology.
>>109470698one less based anon and one faggot like you left behind
>>109471431Oh no! No more text rendered on a static page
>>109470635At this point, (you) have a higher chance of vanishing tomorrow than LLMs do lolJust accept that they aren't going anywhere and are integrated into our society now.
>>109470698fpbp>>109471411the mouth breathing, contrarian, sexless, over-40 boomers on this wasteland of a board think that if the LLM can't one-shot everything, it's "bad" at coding.
>>109470635LLM's just take humans out the picture. obviously without them, companies can just go back to hiring humans again.
>>109474802Except humans are far more expensive, lazier and even a liability to look after.With AI I can ask it to draw me a picture a thousand times and it will never complain. Now compare that to hiring an artist who might bitch and throw a temper tantrum if I asked for even a little change?Life is better now. I do not want to go back to a pre-AI world.
>>109474055It's amazing how many retards actually think they can make AI/LLMs go away.Even if OpenAI went bankrupt they could just release their models for free over the internet as one final "fuck you". Now anyone can download it on their home computer.But the luddites were never smart enough to see that...
>>109470635it'd not, suddenly less slop and lots of roadmaps pushed.investors seething.
>>109470635 Incorrect the economy is being propped up by the AI investments disappeared would cause a stock market crash
>>109474948If OpenAI goes bankrupt, which will probably never happen, they will just get swallowed up and replaced by any of the other major Ai companies. This shit is literally never going to stop lolThere's a reason why so many scientists and engineers that have worked on these things have quit their jobs and retired early to enjoy the rest of their time.
because technology within a technology is almost never revolutionarythe internet is revolutionary, but nothing/no app or site inside of it is revolutionary. social media sites are big, but not revolutionaryright now, llms primarily build and enhance apps or sites. thats what all of the benchmarks focus on and the majority of the post training efforts go into it
>>109471362>my word guesser guessed correctly once that means it always guesses correctly and I can just use all results without verifying themyou should be fired for criminal negligence
>>109470635Infrastructure being built will help the eventual AGI, that's a big deal. But for the most part I agree. It's not that important. The hype was a lie and continues to be a lie.>>109473143This. Sloppers have no comprehension of what actual programmers can do though because well... you know. Obviously. They have a massive fetish for imagining that people are being "replaced" and losing their jobs. Their jealousy is astounding.
>>109470635i cried when the terminator went into the molten steel
I would have to ask people here or google random questions instead of le AI which sucks because the people here are bitter retards and if you google something you probably won't get an answer to your ultra specific question you'll just find a reddit thread with zero replies. other than that idc.
>>109473220The problem is never just simply improving something, but instead in how to do it in a series of small incremental steps in such a way you deliver something you can measure every step you take. Big bang changes are always a risk.
>>109470698A decade of what is now considered 'lost research' in algorithmic information theory and the ability to get around the geometrically expanding interconnect bottleneck latency issue hyperscalers face as they scale up (the true bottleneck to AGI). It doesn't help that I alone possess this knowledge and knowhow.
