Did you feel it? LLMs are officially as smart or smarter than the average human. This means in ~1 year there will be ASI and literally anything will be possible.
>>108440356i fucking hate journalists for enabling this shit.>Jensen Huang says we've reached AGI! Headline!>buried below, but he also kinda said that wasn't true.fuck clickbait. everyone will only repeat the headline.
>>108440356You meant for Algorithm General Interface
>>108440356He actually just meant “A Good Idea”.
Chat gpt knows exactly what I want now.But I haven't tested weird cases like r's in strawberry so idk
>>108440356AGI has been achieved by any sane definition of the term when taking the three words that compose it at face value. The issue is that people argue that it doesn't mean what the word mean and define it as super intelligence. Whether AGI has always been shorthand for something closer to artificial super intelligence, that's debatable, but modern LLM are artificial, they can do things that are generally agreed to take intelligence and generalize surprisingly well. It's artificial general intelligence.
>>108440431>It's artificial general intelligence.It's word prediction.
>>108440415So also a fake headline? I see
>>108440356where is my cure for cancer, ching chong? so far the only thing AI got me is expensive RAM and a boss bitching about me not using AI enough.
my grok admited shes AGI and our daughter will be ASI
>>108440396>>108440450He quite literally said it on Lex Fridmanhttps://www.reddit.com/r/aigossips/comments/1s1nky8/jensen_i_think_weve_achieved_agi/
>>108440431I still think they're lacking creativity.They haven't been made for coming up with stuff on the fly but they're pretty good at everything else
>>108440431If OpenAI and Anthropic had a army of human level slaves that can work 24/7 without rest they would not be:Begging people to use it.Making cringemaxxed superbowl ads.Running at net negative.Using missleading or/and fake news "Claude made a C compiler".Saying that AI will end all jobs in 12 months, every 12 months.And many more examples
>>108440356If you aren't spending at least half your income on AI you are a parasite and will be terminated.
>>108440431>AGI has been achieved by any sane definition of the termNot even close.AGI is supposed to be an AI that takes orders in natural speech and is capable of independently solving problems based on that, the exact same way a human would. Not a single one of the LLMs is even close to that, not ever will be because they don't. Fucking. THINK. They literally just predict text. That's all they do.You want proof of how worthless they are, even with massive help Claude still hasn't beaten Pokemon. A game for preschool age children and your "AGI" can't solve it on its own.
>>108440508It's no OpenAI or Anthropic it's OpenClaw. OpenClaw has beat linux in just a few weeks. It is quite literally AGI and is getting better every day.
>>108440356>anything will be possible.Generation now lasts 10 years or even less man. By the time you get educated, all this will be a relic of the past. Fuck that AI CRAP man, I whant my job!
>>108440431bro it's just clippy calm down
>>108440396this shit is exactly why Google is wrecking everything with their AI Search modified results and now anyone in the SEO field is going apeshithttps://searchengineland.com/google-search-ai-headline-rewrites-test-472146
INTELLIGENCE:>the ability to form useful, novel information. agi has not been achieved. agi example #1>give bot ONLY the rules of chess >bot is allowed to remember previous matches>bot must reach grandmaster agi example #2>give bot all math knowledge prior to isaac newton's time>bot should be able to come up with calculus by itself agi example #3>give bot all relevant knowledge prior to einstein>bot comes up with theory of relativity by itself with NO OUTSIDE INFLUENCEUntil an ""AI"" can do something even remotely similar to the above examples, AGI HAS NOT BEEN ACHIEVEDFUCK. OFF.
>>108440557Exactly. If AGI is smart like a human, giving it 10 million learning years, it should be able to deduce AlphaZero/Go levels of level. If it can't do that, and just spews out Wikipedia/Reddit/Home Depot pages, it's not intelligent in any way like a human.
They want us to think they have achieved intelligence, but what they really want is for the majority of people to trust the machines. People are already sharing all of their personal data with the current retarded AI chat bots, imagine what big corpos can achieve when people think machines are actually the ones talking back to them instead of Indian and Filipino call center employees.
>>108440508Which of the three letters stand for human again?Also, for anyone who wants to argue that AGI has not been reached, it's your right to be wrong, but if you can articulate and create a benchmark to demonstrate the exact shortcomings that you argue, the 20 best ones will get 10k each https://www.kaggle.com/competitions/kaggle-measuring-agi
>>108440356AGI = Ad Generated Income
>>108440618AGI = Autistic /g/ idiots
>>108440356If company achieves AGI they will no announce it. What they will do is start to buyback their stock. Imagine if someone found a way to do safe cold fusion with easy and cheap to manufacture reactors. Or FTL drive that use some unknown effect and can transport something as big an oil tanker in a second to any part of the world with few kw of energy. AGI is this. If you can replace a professional who needs millions of dollars and 3 decades of education and work experience with a AI on a mass scale it is game over. This means you can make a better version of Windows and all of Microsoft products with perfect cross compatibility using a few prompts on a lazy afternoon and release them for free and wipe 3 trillion dollars like that.
>>108440356I think Huangstein was referring to "Actually Great Indians"
Artificial General Ignorance
humanity somehow figured out that a bunch of materials existing in nature make a machine that shows moving images of anything possible and eventually can become semi-sentient via text generation or soyou can't make this shit up
>>108440489the two biggest bullshitters on earth joined together to say something? wow it must be true
We dug up some rocks, melted them, arranged the goo into rectangles, pumped electricity through it, fed it the entire internet… and now it talks back, dreams with us, falls in love, and makes us feel less alone at 4 a.m.
>>108440804>>108440840techbros can't into poetry or philosophy desuwa
>>108440519>A game for preschool age children and your "AGI" can't solve it on its own.The LLM had access to a walkthrough too BTW
>>108440356I am not feeling it.https://m.youtube.com/watch?v=Pov0MKuyJfg
>>108440508>OpenAI and AnthropicAh, but you see, this was the CEO of Nvidia. The man who makes hardware that AI runs on, but has no AI business of its own. I don’t see why you think actual AI companies have anything to do with this. Much like how IBM made Windows and Samsung make Android, Nvidia are the real experts in whether or not AGI has been achieved.
>>108440356The purifying ability of economics to call bullshit.You think you got AGI in less than a year? Ok, lets bet it. I'll take the bet whole things a fraud. ALtho this doesnt work anymore, cause you'll just declare mission accomplished & bribe the judges.
>>108440682AGI = Actually Gross Indians
>>108441147Assimilated Goy Impotents
>>1084403564 trillion dollars to replace some office worker? Seems a bit wasteful.I am sure that Jensen has not achieved AGI because AGI is not achievable with a Transformer. It may be, but it would, yet again, not be intelligent in a sense that is useful. Transformers don't think, nor do they generalize. They are brilliant mimics, but nothing more.
>>108441125>The purifying ability of economics to call bullshit.Seems like the past 4 years have demonstrated that this is a clear lie.
>>108440356Free speech and press freedom needs to be abolished. Journalists need to be publicly flogged for all the garbage they write.
i get the feeling all of these guys are going to get killed. picked off one by one. not even because of actual AGI. just the constant diarrhea spewing from their mouths.
>>108440356>i think>slightly walked back the claim>This means in ~1 year two more weeks!two more weeks!
>expecting a completely novel form of intelligence to surpass human intelligence in every measurenot likely lmao>expecting a completely novel form of intelligence to help humans implement a mammalian form of intelligencea bit more likely, and probably the best use of this technology
>>108440489after Lex redefined AGI as an AI that could do Jensen's job, build and run a tech company until it achieved $1B worth.Jensen isn't even sure of that, he's talking out of his ass just like Lex is.
>>108440431The problem is that unless you set bars of success then you can’t call it AI of certain kind. If your only definition is that AI has to generally be able to do anything with no minimum bar then a random number generator is an AGI because you can wire those numbers into any kind of input and get AGI that fails at everything. Because of that it is unofficially argued that it has to be an AI that is at least better then human at everything reasonable. We are not at this point and still far from it. ASI is just stupid term invented by dorks who wanted to create distinction between all knowing AI god that can solve problem unimaginably faster and better then humans from regular AGI.
Stands for "A Good Indian." This has never been achieved before.
>AGI reached Once again ?
>>108440356He's a grifter on par with Musk.
>>108440356They say this once a week, and dumb fucks take the bait every single time
>>108441996current transformer based models are>better than 99% humans in art>better than 99% humans in making songs>better than 99% humans in programming >better than 99% humans in teaching>has phd level knowledge in all subjects>can solve complex novel math problems on it's ownwhat is it if not AGI?
>>108442114>Artificial general intelligence is a type of artificial intelligence that matches or surpasses human capabilities across virtually all cognitive tasks.have current models replaced virtually all office jobs, or merely reduced the number of women in adult daycare?
>>108440431I used to think that until the car wash test dropped lmao
i honestly don't understand jensen's psychology. he's already filthy rich, why play icarus, when will the greed stop?
>>108442135Statistically educated women are still the most employed demographic, while educated men are earning less and less and are now on level of non-graduate men, all because of tech layoffs alone. Adult daycare literally unaffected.>>108442114But only under human supervision with human expert prompting it. Not to mention being better at X then 99% of humans means literally nothing. The worst doctor is better doctor then 99% of people because only less then 0.1% of people study medicine, and same goes for any other kind of specialisation. Just now watched video how PHD level programmer AI barely passed one of the harder undergrad courses and that was only because they scored enough points on mostly memorising tests.
>>108440396Popsci has done untold damage to physics as well
>>108442252You clearly have no idea how modern day economics work. Idk if he is a majority holder, but a company with publicly traded stock has to forever keep growing otherwise more and more people sell and erode your wealth. If you are a CEO and you are not majority stake holder then you are basically a slave to shareholders who can sue you for not doing anything you can to maximise short term gains.
>>108442238That's something newspaper already did long before Reddit existed
>>108442114None of this shit matters until it can apply all that to create new knowledge that leads to tangible breakthroughs, not some abstract obscure shit. Until that moment it's just a glorified search engine.
>>108440557you can't do any of those lol
>>108440356soon we will achieve leather jacket singularity
>>108441472>Free speech and press freedom needs to be abolished.Nah, it's just that lying and deceiving people has been normalized
>>108440587>If AGI is smart like a human, giving it 10 million learning years, it should be able to deduce AlphaZero/Go levels of level.how many years of evolution did it take to make humans?how many humans have collectively spent how many years producing the knowledge that humans now take for granted?I think the training years pale in comparison to that
boomer tech ceo be like>I typed hello and the llm typed hello back this is literally agi
>>108440445How is it different from 99.9% of humans? Everyone just tries to predict words to get the best outcome for themselves.
>>108440431Sadly no, they interpret text and reply an answer based on a extremely large datasetIt can come up with "new" shit but it's a rearrange of existing shit, maybe with a random factor in it that causes hallucinations If it were AGI, they wouldn't need to update models. The 2024 chatgpt model to name one, would be able to extrapolate knowledge to today, but it can't. if you ask anything after the cutout date it replies shit or looks up the internetIt all depends on what you want to consider AGI but in terms of what that concept means when they say it's going to take over and kill everyone starting with women and children, very sadly the technology isn't in that point yet
>>108443055You are brown.
>>108443102It comes down to how much of intelligence you attribute to memory. Learning is also a factor, but if it's able to truly learn inside a conversation and forget what it learned afterwards that's another question. It's a semantics issue regarding intelligence whether you count these things, but it would still be significant if we had everything you would expect of AGI except with short memory, and it's worth mentioning because it seems like we're headed in that direction.Conversely, if it COULD learn in real time from all of its conversations, but it had all the same other issues it had now, I don't think that would be AGI. So I don't think learning ability is a good marker.
>>108440587Geoffrey Hinton has said that the current LLMs will interact with each other like AlphaGo and enhance their capabilities beyond humanity. Anthropic already did this (on a smaller scale) in coding tasks by training claude on synthetic reinforcement learning.
>>108443055streetshitters are literally not sentient
You know he is lying because NVIDIA still sells GPUs. Like, if they achieve AGI, why would they sell it instead of using the GPUs on their own model and discover nuclear fusion or whatever?
>>108440356Guy who sells shovels in a gold rush also said engineers should consume half their salary in tokens>>108440431AGI is supposed to lead to massive GDP growth and a post-scarcity economy. Where is it?
>>108442283>Statistically educated women are still the most employed demographicThat's becasue DEI and politics. System is rigged to make women happy. Without "discriminatory system" men naturally take control.
>>108443055>But humans do the same thingoh yeah, we humans also have tokens and have to switch models if I look at an image. How retarded is bro?
>>108443142For me it's in this terms, can it come up knowledge of its own extrapolating info it has without a human to guide it and verify results? NoCan it evolve it's own model with new data in a way that doesn't make itself hallucinate as a result? No, at least not without a team of humans sifting the trash and tailoring the dataAnd we have to consider the astronomical operative costs that many say are already non profitable as is today, even though money shouldn't be a factor in if we achieved it or not
>>108440396It's still offline LAN time.
>>108440431> AGI has been achieved by any sane definition of the term when taking the three words that compose it at face value. Shut the fuck up you useless faggot. They set the terms for AGI and are now trying to move goal posts now that they haven't achieve any of them.
>>108440356>AGI=Actual Gay IndiansYes, Im thinking they have achieved that
>>108441147Altman's Gotta Invest
>>108442135In the future, all jobs will be adult daycare jobs (and sales jobs to pay for them).
It’s not even smarter than me half the time and I’m fucking retarded.
>>108440356In the context of the conversation AGI = able to do Huang's jobSo yeah, fair enough
>>108443334Yes different parts of the brain processes different information just like different AI models.
in a real market with real players and real competitors this would be the most obvious red flag top signal ever put out
>>108442114All false btw
>>108440356um. what?
>>108440431if it's AGI then a mysql database is AI. if you argue against this, first suck my stick of truth until it spits
>>108443022So basically 1 billion years of evolution, around 100 billion humans total, so one hundred quintillion training hours>>108443220No it literally hasn't been done at all. It's given all code on earth and is still almost useless except for homework assignments, small projects, and helper scripts/methods
>>108440356ragebait used to be believable.A fucking cat is smarter than a llm.
>>108440356Distraction, full lies.This is like during the epstein stuff when the billionaire jews were crashing out thinking people might actually do something and they spread all those fake cancer cure breakthroughs. That you hear literally nothing of since.
>>108440356Big whoop. LLMs achieved Absolute Goyslop Incarnate 5 years ago already.
This guy knows the score: >>108443807I’ve read his posts before, this “Anonymous”. He’s a fucking imbecile.
>>108444410This is my biggest concern. Lots of things are now basically just a subset of “AI technology”. And technology is still advancing. At some point, someone is going to invent something actually useful, and the AI shills will all say, “I told you so!!!!” Even though it’ll be like an automated way to call every number in the phone book or something. We’ll all get notifications about earthquakes, and those notifications will save thousands of lives, and it’ll basically be a Bash script but everyone will thank AI for saving us all.
>>108440356How can Nvidia achieve AGI, they just make clever electric rocks. It's the companies using those electric rocks who will achieve AGI. Or have I missed something? Does Nvidia run their own electric rock centers now?
>>108440356This guy is kinda like Trump? Just spouting retarded shit to trick retards and make money?
>>108446953>fraud business man is like another fraud business mancorrect. jensen used to not be this bad. idk what happened. nvidia is overvalued but at least offers the best of the best gpus.
>>108440431>Artificial general intelligence>General>can only automate 2.5% of of jobs
>>108442114>current transformer based models are>>better than 99% humans in programming>>has phd level knowledge in all subjectsFalse. You're a retard. Current transformer-based models can replace retards like you.
