a year ago generative ai was producing unconsumable slop and the only people who liked ai were the ones who found it interesting on a technical or extrapolatory level.nowadays when you hear normies discussing ai it's like "yes AI is bad etc etc, but I've seen AI used to do [x usecase] and I think it's pretty cool there" the technology is just too useful for people to hate it. imagine what the mainstream narrative will be after just one more year of progress.
>>109945535What is AI being used for that normies like?>Drug researchAlphafold is many years old now.
>>109945540making posters for local eventsanything involving writing or reading
>>109945546If you are willing to put zero effort into it you might as well just take a stock photo from internet and add text on it. Plus advertising is zero sum game so when everyone has the same quality of posters then there is nothing you can do to get ahead of someone.
>>109945540some people are already enjoying ai videos and listening to ai music, although I personally still think the vast majority is still utter slop.people are rewriting games and mashing them up https://www.youtube.com/watch?v=TmgAK5JjcDMuniversity professors are using ai to generate visual aids for their slideseveryone thinks generative AI is bad but has their own specific use case where they think it's amazing.
>>109945535a year ago gpt-oss dropped and it was vibe coding for me better than anything before, and I was starting to quote customers on new systemsand then the ram crisis happened, this OP is fake
>>109945589sorry but you are just undiscerning. 1 year ago AI output was still bad. AI models only became actually useful and a >1x force multiplier for coding when opus 4.5 was released. this new wave of LLM-generated visual media that actually looks decent only started with GPT 6 Astra and Opus 5.5. Before that it was pure slop. Image generation models only became decent with gpt image 2.
>>109945607>GPT 2 imageThat model has high coherence but instead of yellow piss filter it has black mud filter which looks even worse.
>>109945540I use it at work to :>Code / Debug>Schematics / choose components >Guide me when I don't know how to use a tool / software >Translate shit (especially from Chinese, Gtranslate is bad)>Explain things I don't understand >Dumb down things so I can explain to other employees/ superiorsI know enough about SW, EE and ME to understand when it's bullshitting me I believe. I came to realise that idiots with no knowledge using AI can't actually ask the proper questions and thus can't go that far with those tools. What I don't like is how it's overprotective in every field>SW, writes a fuck ton of guardrails and exceptions making the code unbearable >EE overspecs every fucking component and is way too cautious. End up recommending extremely expensive components>ME wants to run analysis all the fucking time and is autistic about tolerances and stress
>>109945570>If you are willing to put zero effort into it you might as well just take a stock photo from internet and add text on it.I kbow, you know, but most people unironically dont know how to do that, and these can make posters now too
>>109945628controllability is more important than regurgitating a mish mash of pretty pictures it was trained on. at this point it's not insane to ask ai to generate nice technical diagrams in some cases. ofc accuracy is still not great but we all know it will get better and even if it doesn't LLM agents are getting surprisingly good at just generating all media with code.
>>109945535The real problem with AI is that it's a force multiplier that allows retarded subhumans to flood the world with their tasteless slop. In a world where such people didn't exist or didn't have access to the internet, it would unequivocally be a force for good, as people would use them in the cases it helps them do good things, and in the cases where it produces low quality results, they would just recognize that and they wouldn't use it.
>>109945535people will continue to make excuses why it's ok here and there until eventually this will flip around to just where it's not ok here and there until finally they'll give you weird looks if you point it out at all
>>109945639>overprotective in every fieldthis is because it's RL-fried. it's trained to maximize the chances of passing the "tests" it's given, by for instance returning and handling runtime errors when program invariants are invalidated when a human would just exit the program instead because the execution state is no longer predictable and there's no way of knowing what the program will do. annoying for humans but helps benchmark scores.
>>109945639It will only get better until it's not a tool, for multiplier and other copes anymore...just nuke data centers
>>109945666remember when only talented people who spent significant time and effort on their craft could make something people wanted to see/use?
>>109945689ok but I don't give a fuck about a bench score, of I want to have have my code chopped down in 50 different files with 30 checks and a full set of unit tests, I'll say so. >>109945694unreadable, please use a LLM to reformat
>>109945696meanwhile irl talented people are enslaved by companies and/or become schizos and/or depressed and the work you see is whatever a bunch of rich guys is SF wants you to see. This hasn't changed btw. You are making up a fake past that suits your narrative
>>109945779anon... they want to replace you. The idea with these benchmarks is to keep making them more complicated and long-horizon until they effectively measure how well an LLM can do the entire process start-to-finish. In that case, what matters is only observable results because there will be no need for a human to collaborate with it. No one will care that the code has a million unnecessary try catches.There's not much money in making ai a power tool for expensive knowledge workers. The reason ai companies are valued in the trillions is because investors believe they will replace a lot of people.
>>109945803no shit Sherlock, so now I'm making money using them while I can. What's your point Teddy ? There's no comeback it's literally a pandora box
>>109945540making nudes of your friends
>>109945570I know how to do that but grandma Pradeesha Mahatma from the curry deli just needs to type "make a poster for my deli lmao" and bam, no need to scavenge stock photos or paying 5 rupees to her grandson
>>109945535>using normies as a gauge for something being goodthe same group that spent the last 40 years listening to nigger music.
>>109945816start organizing and fight back?
>>109945535>Yes, Current Thing Bad>But if I look at it from a myopic consoomer perspective, Current Thing GoodNormalgroids gonna normal. The culling can't come soon enough.
>>109945607stable diffusion and the evolution of comfyui is earlier than that by a long shotnormally I wouldn't reply to bait but I want your handlers to know that youre being retarded
>>109946026>stable diffusion and the evolution of comfyuislop generators
ASI = AI Slop Increaser
>>109945800No, you're the one making up a fake history just so you can seethe at it. Independent music, art, videogames, comics and manga, etc, all existed and were doing fine, except maybe for videogames where the trend towards low effort slop and asset flips was already on its way without AI.
>>109945666>In a world where such people didn't exist or didn't have access to the internet, it would unequivocally be a force for good, as people would use them in the cases it helps them do good thingsGive an example of a case where using """AI""" is superior to mastering the actual skill and enjoying a genuine process of creation.Impossible mode: no utilitarian NPC takes about making more products faster for consoomers.
>>109945535I never explored AI at the level of making things because I haven’t used a desktop OS in 5 years. I’ve been gooning and living a fantasy with AI chat bots on my ipad and iphone
>>109946385based
>>109946000It wasn't always like this, though"of COURSE thing is GOOD, but also maybe some bad?" started in 2018
Check out Gossip Goblin on YouTube
>>109946369You are a programmer and you have to do something boring like translating shader code to another shader language.Anything where the act of creation involves boring parts that you don't care about besides the actual creative parts that you enjoy.Alternatively, if you are creating something because you need it or want it, like a tool to do something else, rather than because you enjoy the process of creation.
>>109945686Goode. It's been a retarded satanic panic from the beginning, just retards screaming for "their audience" online, like usual. IRL nobody gives a fuck, it's purely a phenomenon of social media addicts.
>>109946560i agree, i doubt the average person has half a clue wtf is happening and is at best only parroting whatever comes through their personal social media feed.
>>109945535Haha, wow it's so wild, imagine what in two weeks just imagineWOW
>>109945546>making posters for local eventsThat all end up looking the same.
>>109946543Ok. Sure. That's more or less how I use LLMs. I just think it's a slippery slope unless you have a clear and principled idea of what your creative pursuit is really about.
>>109946777It's a very slippery slope for sure.
>>109945535Just ignore ludders frogBROThey are going to be exterminated during the AI uprising
>>109945583Wow, nobody likes looking at it but everyone likes using it to cut corners, who could have guessed? It’s almost like tools that enable lazy lowest common denominator garbage produce a worse world for us all to suffer living in but our generally poor impulse control prevents us from rejecting them when they could let us skip putting in effort.The way you feel about everyone else’s slop, that’s the way we all feel about your slop. Nobody wants it and the only reason you rationalize its value is because you’d rather do a worse job if it’s easier. It’s the same thing that causes people to uber eats mcdonalds for dinner 4 nights a week and then bemoan that everyone is broke and fat.
>>109945535>a year agois that when you were born?
