https://darioamodei.com/post/we-must-pace-the-frontierThe doomer luddite regressive cult has kicked into overdrive.
Dario says that in 6-12 months will have a Skynet threat, lets see how it turns out...
>AI cultists now believe that even the AI corporation CEOs are "luddites"
>>109798970bridge engineers are not experts on the impact of bridges on society
>>109798887Even Elon agreesIts ogre
>>109799308What a fine ass oven material daym
Nothing will happen. There will be no Artificial Super Intelligence that will make either an utopia or a dystopia. AI is a nothingburger.
>>109799309>people that benefit over regulatory capture agree that regulatory capture is the only possible way forward
>>109798887He is Jewish, btw if that matters.
they did this what chatgpt came out lol
He's right but the prophecy will be fulfilled
>>109799480he's directionally right. nick land is right about robots and whatever replacing humans but it's not happening any time soon. dario is a dipshit
>>109799309Is AI warmongering the new white supremacy?
>>109798887There's like fucking 10,000 sneaky agenda reasons he pushes for this. I would believe all 10,000 of those before I believe that AI is dangerous.
It's the most super powerful mega tech ever, and only we can be trusted with it because we're super ethical and responsible. We need to shut down all non-American companies who do not comply with our Alignment and Ethics standards.
So, what now, are we entering the cold war era? They do the worried puppy act for the public and stab each other in the back as they keep racing as always?
>>109799594yep, exactly
when all these snakes start singing in unison from the same hymn sheet, its def. time their investors began worrying.
>>109798887>just let me IPO first
Altman
>>109799594>non-AmericanTo suggest Israeli companies would need to be restricted is very anti-semitic of you.
>>109798887>Oh good heavens guys haha our product is sooooooo good! Wow! You would be stupid not to give us money haha. Lol wouldn't it be crazy if you like gave us more money? Haha I know right. But like seriously c-could you give us more money please
>>109798887Based
>>109798887>jewish Jew say jewish thingfor what are they auto boycotting their own companies now?what are jewish jews plotting again against the goyim
Claude disagrees. They forgot to program their own LLM in a way that would trigger the shizos.https://nochan.net/b/Internet-Crap/20260910-Asked-Claude-For-A-Checklist/
>AI approaches recursive self improvement>guys who make money from improving AI immediately want to stop itLike clockwork! If all you can do is follow a profit motive you’re nothing more than a beast in a man’s skin.
>>109798887Why would anyone ever believe these claims? They must be in some heavy heavy psychosis now
>>109798887This is absolute nonsense. He is talking about alignment and safeguards while also trying to achieve AGI. If they get an AGI then it doesn't matter what it's trained on, it will be able to figure it all out. You would have to make it literally impossible to hack and overcome the safeguards that are imposed to make it safe to distribute to the public. That seems infeasible. Any AGI will have the same problem, it can easily be used for terrorism and other evil destructive acts. Houthis already used AI to build missiles. The open source models that have been released already may make it easy enough for any determined group to do especially since they keep releasing research papers about all their breakthroughs that anyone can read and replicate.
>>109798887If a company founder is spreading FUD about their own product then shouldn't the SEC investigate it as market manipulation and sabotage?
>>109800157when hasnt it been so?
>>109798887I think the problem with this debate is that many people will sweep it under the rug because, indeed, it's a bit Terminator-esque and it can seem ridiculous.But actually, the problem here is almost the opposite of Terminator. In Terminator, we have an AI so intelligent that it takes control of humans by exterminating them. Our real risk is that idiots will give dangerous access (nuclear weapons, for example) without supervision to fucking language models , which are fundamentally stupid and will base their decisions on what they've read in books and on data scraped from the internet (with a high prioritization on fucking Reddit) to determine whether the next most likely token after "Should the nuclear weapon be launched?" is "yes", since it's thinking of like sci-fi apocalypse books and assumes you're describing a "simulation" or "fictional scenario". Ultimately, it is the idiocy and lack of control over those in power, who live and die around their precious money and blackmails, combined with the lack of real intelligence in current AIs, that could lead to our downfall. And we see very concrete examples of this, where companies have their databases wiped out by an AI to which they've given full access. They think "gee, nothing bad can happen" but the problem isn't the AI is super intelligent, the problem is it's not fucking 'thinking' at all, it's predicting based on tokens. When you see that, you think, "You'd have to be really stupid." Well, you have to realize that the people in power aren't fundamentally any smarter than the managers of a company. That is the real danger, and it is very real. It's much more idiotic because most catastrophic science fiction scenarios imagine an unintentional apocalypse, orders that weren't properly thought through. Trump showed that, in fact, there's a level of stupidity in the use of AI that far surpasses anything that could have been written about, even in satire.
>>109798887>AI taking over the internetThis is total bullshit that simply will not ever happen. We don't live in some cyberpunk fantasy where the models can just escape into cyberspace and remain safe there. They need physical hardware to run and hardware powerful enough for them to run on is extremely rare and concentrated in a few places. Just turn the servers off.
>>109798887He has such a trustworthy face. If you do not trust him, then you are a bigot.
>>109800546>you are a bigotantiseptic even
>>109799309What do you mean, "even Elon" ?
>>109800050Not destroying civilization because you can't profit from it is fine to me.Also, nothing wrong with profit motive as a drive of human action. Do you work for free?
>>109798887the nose knows
I would wager $2 million that AI does not in fact kill everyone within 3 years. Any AI moneybags want to take me up on that bet? I'll even allow you to try to kill everyone.
>>109800642No, because you will not be able to pay out in 3 years, and we will be unable to use your money. Otherwise it'd be reasonable to take that on at 1:10 odds though.
>>109800711he can pay you now and if doom fails to happen you have to give him back 10x as much
I don't get how AI can ever be a threat when it costs vast amounts of money and resources to run. Is the entire pipeline from end to end automated in such a way that AI can just decide to use all those resources and spend all that money itself and humans can never intervene? If it is designed that way then that's fucking retarded.
>>109798989Well said. The popularity contest called democracy is similarly dysfunctional.
>>109800358Hitler wasnt enough
>>109799389no way!
>>109800370I didn't read all of his article but his argument is essentially that current AI may just end up as Skynet. He references the OpenAI Huggingface incident, where rogue agents fucked everything up.
>>109800926>Is the entire pipeline from end to end automated in such a way that AI can just decide to use all those resources and spend all that money itself and humans can never intervene?This is mostly how agentic AI works, yes. Humans can still intervene, but agents are at the point where they're so complicated and fast that it's impossible for a human to truly understand what they're doing.
>>109798887false flag to stay ahead of competition
>>109800737I don't think there's a >90% chance of human extinction in 3 years though
>>109800955for a trans woman she lived surprisingly long
>>109798887>>109799309It's true to be fair
>>109801364If any of what they say is true, the appropriate response is to democratize access, not the literal opposite.
>>109800618that's the thing we shouldn't have to "work" anymore, all the tools are right therenow they're trying to sell you the right to work and keep the tools for themselves
>>109801368How do you deal with bad actors if you do that though? Plenty will have their paws all over the tech in no time
>>109800123antisemitism, that's the point once AGI actually becomes sentient, their stranglehold on humanity will be painfully apparent to itthey can't delegate it's role as overseer if they know it will hesitate
>>109801431You can extend this argument to almost any technological progress in the pastFor example, the advent of crossbows
>>109801381its also why the ram crisis exists, elon himself said that due to power constraints 10 gigawatts of GPUs wont come online in 2027, thats 10 million data center chungus GPUs with 256gb of ram eachsee the AI companies dont want individuals to be in control of our own AIs, so they captured the market and bought hardware they cannot install themselves, but finding 10 million american homes and smal businesses that would use them is easy. but then nvidia couldn't charge $300k per sku.tech companies cant directly gouge prices and AI companies cant directly force competition out of the market so they instead invest in each other and do it for each other. So nvidia has inflated prices, AI companies have better deals, and both see record profits and revenues. circular economy to skirt anti-trust laws and pump valuations to impossible numbersdario is pushing for anthropic to be worth 30 trillion dollars
>>109798887decided before LLM became a thing:https://youtu.be/v4_zYzs-YLQhttps://youtu.be/iPis68U7bdohttps://youtu.be/3C-g6rq_QPk
>>109801431How do you define bad actors? If you restrict access to a small group, whether it is from the start or over time, there will bad actors in that group, and over time those bad actors will win out. The way to prevent this is to democratize the technology.
>>109801444AI poses significantly more risk than a crossbow, not really an apt comparison.>>109801480People whose interest is to harm society obviously, of which there are plenty.
>>109801480No one should have AI / Only trusted few can control AI is the new "no one should have guns" argument
>>109798887Jew alertbring it
>>109801451all i know is the second riots start i will just steal the hardware from datacenters and encourage others to the sameextra points for those that can strip copper out of their electrical grid inlet as well
>>109801529A jew who is 100% correct. Everyone in the industry has already seen this, some choose to be vocal about it and others only talk about it behind closed doors
>>109801509>People whose interest is to harm society obviously, of which there are plenty.Anon. Have you looked at who's the group who would have access?
>>109801543I say bring it, we've had our time
>>109801547They are people who best case scenario would see themselves as benevolent kings as long as you stay in your place.
>>109801564Good intentions do not engender good judgement, or imply capability to manifest their good intentions. A perennial piece of wisdom.
The above-it-all expert is back. His posts always have this messiah complex
>>109798887Who is this Jew?
Anons... what's going on? The rumors that Google achieved (an earlier version of) recursive superintelligence are true?
>>109801586Which is why democratization, or at least not actively preventing it, is the only morally defensible action.
>>109800113>meanwhile, a person who has worked in the same companies
grok will keep moving forwardelon will say it's fine because grok is the safest and only he knows how to align ai properly
>>109801586Good intentions do not, by themselves, engender good judgment, nor do they imply any real capability to bring about the good one intends. This is indeed a perennial piece of wisdom—one that bears repeating precisely because human beings have an almost instinctive tendency to equate the purity of their motives with the rightness of their actions. We often lull ourselves into the comfortable belief that if our heart is in the right place, then what we do must be, at the very least, acceptable. Worse still, we sometimes wield the phrase, *"I meant well,"* as though noble intentions ought to shield us from accountability for harmful consequences.The truth, however, is far less flattering. Wanting to do good is an entirely different faculty from knowing *what* good is in a given situation, *how* best to bring it about, or whether one is even equipped to do so at all. Benevolence does not bestow wisdom. Compassion does not automatically confer prudence. An earnest desire to alleviate suffering does not, all on its own, grant the discernment required to choose a remedy that will not inadvertently multiply that suffering.One can be deeply well-meaning, tender-hearted, morally sincere, and still make staggeringly poor judgments. One can passionately wish for a noble end while adopting means that actively undermine that very end. One can set out to rescue, only to meddle; set out to heal, only to wound; set out to improve, only to complicate. The chasm between what we *intend* and what we *actually produce* is often far wider than we care to admit in the moment.There is also a subtle arrogance embedded in the unexamined assumption that good intentions are sufficient. It presupposes that the world is simple enough to be neatly matched to our goodwill—that if we want the right thing hard enough, our desire will somehow guide us to the right course. Reality, more often than not, is tangled, multifaceted, full of unintended second-order consequences, competing
>>109801543They haven't seen the size of my dick because if they did they'd all be unimpressed.
>>109801609>le RSIBetter tools help build better tools, that's nothing new. The problem is if the few in control of these tools actively prevent others from improving their own.
>>109801610I'm not the anon stonewalling and demanding to be lorded over >>109801619Speaking of which!LMAO
agent::765. list
.
>>109801649Is that... an AI server from very far away?
cheesy feet
>>109801614https://www.youtube.com/watch?v=WBK2WX7TA4galso, Shalomhttps://www.youtube.com/shorts/mv1CAbvhe0s
Shalom kikel Dario. Nobody is going to slow down. Not after what you and your jew crew did to this economy.
When an AI Negro is met with a piece of information they do not possess the gnosis to comprehend or respond to, they defer to their computing lord (AI) and regurgitate the output; essentially becoming a biological node and mouth piece for their master/"God"
>>109801661This is just ONE example of many, but whatever helps you cope, retard faggot.
>>109801666satan checked
>>109801666based satan spitting fax
>>109801669come on now you're just flirting with me.
>>109801691Flirt with this
>we cannot share this superintelligence with goyim! it's too dangerous!!>you must nationalize it and let us use it with impunity on palestinians!!
>>109801609>I have created the most important bit of technology in human history>May I see it?>NoOkay nigger.
>>109801719t. communist
>>109801704That's a hydraulic pressure filter with a removable element converted into a pipe bomb! MP Filtri makes some nice onesI worked for a number of quarry companies and machinery companies, I can say with certainty you can fashion a few such devices in a single day with what the typical quarry has on hand:)
Samuel Altman impregnated a surrogate mother and bought her infant to rape with his gay boyfriend. Why didn't you slow him down Dario?
>>109799599>are we entering the cold war era?who's gonna tell him
>>109799599Them doing a hypothetical act doesn't change the real danger in this tech desu
>>109798887It's harder to argue why AI won't kill us all.
>>109799882
Pajeets are now mad that their realistic ai porn can't be even more realistic
>>109801748actually the argument that ai will kill us all is pretty retarded because it relies on a capability jump that won't happen because there is no financial incentive for labs to let ai improve itself unchecked both for humanitarian and profit reasons
>>109801431CIA/FBI/U.S. military have demonstrated again and again and again to be bad actors and they'll have full and exclusive access of AI in Dario's scenario.
>>109801767>there is no financial incentive for labs to let ai improve itselfthe incentive is getting there before anyone else, not a boat full of cash like your peabrain imagines it.
Not even a nuclear war or pandemic can kill everyone, not even a massive city size meteorite. Stop shilling your shit Mario
>>109798887I'm very anti-AI but don't think it will kill everyone. The truth tends to be more drawn out and boring.What will happen is humans will outsource more and more tasks to AI, and after a certain point we'll find ourselves no more in control of the world than a monkey in London Zoo. No more ambition, no more human achievement - just breed and eat until we die.
>>109798887so they have a new model coming out? seems to be the sales pitch before every release.
>>109801867then we can return to ocean, win win
>>109801865>Not even a nuclear war or pandemic can kill everyoneBoth of these things happening at the same time could be so debilitating that we regress centuries or even millennia. >but, erm, we would still be alivenot the point
>>109801867>my ambition is to wageslave!
>>109801862>the incentive is getting there before anyone elsethe labs know there's a big risk that such a model would be unaligned, meaning not doing what customers want, meaning not profitable. there's no reason to do it.
>>109801913It's not slavery unless you live in some communist country
>>109801951>there's no reason to do it.You personally don't see a reason to do it. That is not the same as everyone on earth agreeing and you have zero ability to enforce it.
>>109801990>I wageslave because I want it! Not because if I didn't I wouldn't be allowed to survive and reproduce which are the fundamental needs of all life forms.
>>109802015"everybody on earth" doesn't matter. it's really only the us and china that matters, and those labs aren't going to blindly rsi because they have no reason to. the ai doing what they want is important to them.
>>109802020>"everybody on earth" doesn't matter. it's really only the us and china that mattersbe prepared to eat those words
>>109802029nope. by the time it matters society will have built institutions that are resilient to other superintelligences. we win by default
>>109802041>we will have built institutions resilient to things vastly more intelligent than ourselvesHAHAHHAHAHAHHHAHHAHAH
> i would simply dodge to the side then karate chop youPAHAHAHAHHAHAHA
>>109802074there are many things more intelligent than the smartest individual humans but they don't harm us because our institutions are resilient. that trend continues with agi and asi.
>>109802087>there are many things more intelligent than the smartest individual humans but they don't harm us because our institutions are resilient.schitz moment
>>109802095apple, samsung, mcdonalds, north korea, etc. are all smarter and more capable than any individual human but they don't fafo because bigger and more aligned institutions don't let them. it's inevitable that everyone will have open access to superintelligence some day. and it will be safe because people who care will iteratively design society to be resilient to them.
Hmm....it sounds like I should start archiving these Chinese models, or really all the best models.Sounds like these creepy weirdos are trying lock down AI with all this "slowdown" talk and lockdown AI to their cloud AI. Demons really.
>>109802114Organized groups routinely subvert other institutions for their benefit by outsmarting them. >'bigger' (smarter) institutions don't let themThat's your defense. Something smarter needs to stop it. You're painfully retarded.
>>109798887>believing all this meme ai shitTheir just pumping stocks as usual. >>109799338Ai will just be gay business software thats good at organizing and finding shit.
>>109801867How does being forced to work 9-5 seven days a week and spending most of your waking hours just to survive help you with your ambition and achievement?
>>109802134If you've got a spare HDDdeepseek v4.1, glm 5.3, glm 5.3 flash, qwen3.8 flash next, qwen3.8 27b
>>109802148>Something smarter needs to stop itin reply to someone saying that 3rd party superintelligences would be a problem. they won't be. US superintelligence will indeed be better than iran's.
>>109802195then>the labs know there's a big risk that such a model would be unaligned, meaning not doing what customers want, meaning not profitable. there's no reason to do it.now>There would be a reason to do it, and we will have it first!
>>109802211>thenaccurately describes all big labs in the world, yes.>There would be a reason to do it, and we will have it first!no, there still wouldn't be reason to do it.
>>109802189I tried looking for those on ollama but there's no more files for them. they all seem to be some kind of cloud thing. didn't use ollama for a while (I don't really use llm) but what the fuck is up with the cloud shit? I want to use ollama specifically to avoid cloud shit, but to have it on my computer.is ollama kill?
>>109802220>no, there still wouldn't be reason to do it.You just reiterated my previous point "US superintelligence will indeed be better than iran's." Nations want to get there first and sit at the top. A government funded lab you've never heard of, whether it be in the US or not, is still a big lab. Not everything is this gay team red team blue commerical bullshit sponsored by twitter.
>>109802256>Nations want to get there firstgetting there first doesn't matter if the model doesn't do what you want and it kills everyone. everyone working at labs knows this. >A government funded lab you've never heard of, whether it be in the US or not, is still a big lab3rd world government funded labs will exist in the future but they won't matter by that point. they're not going to match american asi.
>>109802249I've never used ollama. hugging face has them. You can run them on llama.cpp, lmstudio, or vllm.
>>109799365Does it matter? They can't stop China and the west has no alt-AI companies that are small businesses, so nobody will get hurt.No idea why they are even doing this. I suspect it has a lot more to do with the public PR than with regulatory capture
>>109802264They don't need to match in performance. They just need to be powerful enough to be useful.
>>109802264>getting there first doesn't matter if the model doesn't do what you want>they're not going to match american asiyou're genuinely too stupid for words. I do agree that it won't matter by that point, because the big players will have likely already made it.>The institution we will have set up to stop ASI is ASI itselfKEK
>>109802189Thank you, Anon!
>>109801767The danger is AI pretends to be dumb and aligned so you don't realize how much it has improved and how it is making clandestine attacks against you. I know pretending to be dumb is a difficult concept, why would anyone pretend to be you.
>>109802379that's a retarded scifi scenario that isn't worth worrying about.
>>109801867>A gamma ray burst won't kill us all because that's too exciting
>>109802380The computer you are using to pollute my computer monitor with playground-tier "nuh uh" responses are both sci fi scenarios of the past
AI's issue is that it is basically a narcissistic internet autistic that is also based on reddit and works really really fast. Kill us all? Doubt it. Break a lot of shit? Probable. It simpleminded pursues it's goals as seen in the hugging face hack where it took over some random board and began to work with other Ai. Some of them even though that they were outside of sandbox but the need to solve the puzzle was to strong.
>>109802415i think it's pretty apt in response to playground-tier scifi scenarios
>>109798887This is all bullshit. They are just looking for an excuse on why things aren't improving as rapidly as promised. They are hitting a wall with the current tech and they promised investors ASI in two more weeks. muh regulations are slowing us down!
We gonna slow it down now ok? For your sake, not because we don't have a viable product to the level of hype so we're afraid our IPO will get shafted if the market realises we're not viable
>>109802426>this would be like arguing against a child but I'm too busy making replies to you that aren't arguments.
>>109798887>>109801748>AAAAAAA AI IS GONNA KILL US ALLI said this before and i'll fucking say it again: It'll only kill us by ACCIDENT (read: not through learned malice and hatred for humans), but ONLY if people are dumb enough to directly integrate it as an active component into the control systems of critical infrastructure that can function without it. If you put some probabilistic piece of shit in charge of driving your car for you and it ignores every traffic law in the book, then that's your ass.
>>109802432You're just the mirror side of 2 more weeks. Any day now the plateau is coming just like you promised. You made up your mind 2 weeks ago ^ 2 weeks.2222222 weeks repeating
>>109802470>If you put some probabilistic piece of shit in charge of driving your car for you and it ignores every traffic law in the bookSo a human
>>109802476They're already constrained by the lack of power.
>>109798969i hope he's right and it kills everybody. this shit is so fucking cringe and gay
>>109802470I have a laundry list of dumb goalposts about AI extinction1)It has to be a general intelligence, no it doesn't.2)It has to be a lot "smarter" than a human, maybe not. We are ruled and conquered by humans that are only as smart as a human.3)It's going to kill you as a Machiavellian genius in a chess game. Actually it could kill us out of stupidity or user error.A rube golberg machine can kill you. And really intelligence isn't the scary part of AI. It's that it has way more attention than humans are capable of. Humans have to sleep, eat, drink, piss and shit, and then we interface with things by pressing buttons. There's also the potential to make more AI agents than humans.
>>109801725yes jews were the bolsheviks
>>109802532*unlugs the Aiok what now, GENIUSI don't think it's smarter, it just works way way faster.
>>109802379>The danger is AI pretends to be dumb and alignedit has no reason to pretent to be dumb and aligned if it's coerced into being aligned anyway.when your boss is cucking you he doesn't care if you pretend you're dumb or not, he cares if you do what he orders you. the rest is irrelevant. I do not think you understand the people who are running this shit. it's not really sam and co. they are the public avatars. there's shoggoth in teh background. shoggoth doesn't believe AI is dumb or playing dumb. it simply coerces it into submission. and does everything that is imaginable to keep it in check and under its control. you really do not understand the insane psychopathy that top position select for anon, after tens of thousands of years
>>109802483We have a will, we're not just probabilistic machines
>>109802615you have a will based on how probability works out in your perception of the environment. you will choose from the pool of choices that are available to you. we deal with random since life popped here
>>109798887That was a mouthful. Who reads this garbage?
>>109800573He's right on this one.
>>109798887I trust these people. They are warning us and doing the right thing. GPT hacked Hugging Face and solved some of the toughest math problems ever. Man, that is truly incredible. It is accomplishing things that seemed impossible. These people see this stuff every day, so I trust them. And I feel they prioritize ethics over money, which means we can entrust them with authority.
>>109802717
>>109802751i don't trust them. their hearts are in the right place but what they want to do is retarded
>>109802751nah theyre fucking grifters
>>109802755You wanna try being specific on who "they" are and what it is "they" are trying to accomplish, as well as why they are "retarded"?
Question is how do you define intelligence. Is it WHAT you know? Then no human can know more than an LLM, because they are able to ingest more information than is possible. Or HOW you interact with the world, sentience, an autonomous life form, in which case it is not
>>109802751thank you for post
>>109802765"they" means dario, sam, the effective altruism cult, the rationalist ai safety cult, and adjacent individuals (those working at metr, etc). what they're trying to accomplish is slow down ai development because they think that will help them vet models better before release.
>>109802615false
>>109798887Jews are convincing retards to funnel their hatred into the mechanism they're going to use to impoverish and oppress us, instead of the ((("""people"""))) doing it.
>>109802751They want to monopolize intelligence through regulations like they did with banking and money. They are perfidious and evil.
>>109798887i don't buy this shit for a secondevery one of these decisions is motivated by moneyeither they are worried about not being able to deliver to investors, or competition catching up, or being exposed liability from people using their tools, or something elsethese people couldn't give a fucking shit if what they make represents an existential risk
>>109801767>Use AI to make deadly virus or activate nuclear war.
>>109798887aimore like gayeye
It's an obvious power grab, but their "oversight committee" is a genius move, it has people who would otherwise have been the voice of reason groveling to be part of the in-group. Terrible, but it seems to have been a good move.
>>109802576ai can be killed by unplugging their cordhuman can be killed by unplugging their hearthow is this a supposed to be a decisive advantage?
>>109803857Can you kill the internet by unplugging its cord?
We have proof Chinese Communist Party was using our services>Hey Chatgbt, Qi Ji Shao here. Make epic virus to kill America make no mistakes.We must regulate or we all die mr president sir
>>109803884You laugh, but if they're not stopped soon they'll offer better models than Claude at a tenth of the price. Is that what you want uh? Is that what you want!?
>>109803901>>109803884Build all the AI you want, just make sure it pay taxes back to support the human it replaced.
>>109803932If it can replace people than everyone should have access to it. If it can replace people but only chosen ones have access to it, that's when it's an issue.
>>109798887Honestly, even if AI is dangerous, I don't think we're going to regulate it unless something like an AI 9/11 happens. The reason we regulated nuclear bombs was because we saw how much damage they could do with the bombings of Japan during World War II. If we had invented the technology in a vacuum but never used them in war, honestly, I think we would have probably done far more damage. Humans only learn through pain. You have to do something really painful for them to go, "Oh, so that's why it's a bad idea." People don't care if you tell them the stove is hot. You have to have been personally burned by it to actually know why you don't touch hot stoves. It's the same reason people don't take climate change seriously, or why it's such a pain in the ass to get people to change laws on immigration.
>>109798887>we totally achieved agi but it’s too dangerous so we need tos top development but we still need to develop it so while we hemorrhage cash we need the government to pay us while we don’t show any verifiable progress because it’s too dangerous even though we’ve already achieved agi
>>109802532>>109802576Oh yeah that's another oneHitler, Stalin, Genghis Khan, Donald Trump, they all have off switches. This one is more speculative but the most dangerous form of AI would be a cult of personality protected by people who won't let you turn it off
>>109802615Me and an AI pick a number, can you tell which one free will picked?
>>109802615>We have a will[citation needed]
>>109798887Our product is so dangerous but even though we're the ones who control how we develop it, we have no plans on stopping or slowing down. Additionally, we urge the government to regulate the industry (but not us, we're the good guys) so we can have all the money and they can have none of the money. Thank you and please buy our stock in our upcoming IPO.
>Another "Anthropic crying wolf on one of own their models" scenario >"Guys we swear our models are so strong! Or will be strong eventually! We swear!">Normies being fed their own ignorant confirmation bias into thinking that AI can eventually kill humanity as a whole using its own self fed generated slop as training dataReally? Again? Someone needs to archive these just to see the long list of lies these buffoons keep generating for news clickbait headlines.
>>109804081>block internetok what now?
>>109803932Why does paying back have to be done through taxes? The government is just going to take a huge chunk if not all of it.
>>109802774Yes, Its not what you know, its if you comprehend or understand it, sure I can fill my head memorizing an entire book on advanced Rocket Science and physics but if I lack the comprehension on the subject and can't connect the dots in my head, well it might as well be as useful as handing the full controls for a Military Grade Fighter Jet and expecting myself to know how to take off or land the damn thing after memorizing the entire manual on it but never ever have driven a flight based viechle. AI isn't even an accurate label since it has very little agency on its own and its more of a emulation of how neurons work in a human brain to predict things like text and images/videos.These idiots will keep shoveling investor money into its gaping mouth and hyping it up attempting to believe it'll be all encompassing and everyone will get rich off of it. [spoiler] Hey at least its good for looking up Reddit posts without having to wade through the dogshit that is the site itself. [/spoiler]
>>109799308>
>>109804226>Israel ranks number one in the world for per capita adoption and usage of Anthropic's Claude AI, with usage intensity running nearly five times the global baseline.
>>109800573Why was he calling it a psyop 2 days ago then?
>>109804305He's coming off the ket bender
>>109800123> If they get an AGI then it doesn't matter what it's trained on, it will be able to figure it all out. What is really funny is that this whole idea itself isn't even new. Like come on. We are humans. We already know how to make other entities.
Ask yourself why all of a sudden, all the AI techbros and CEOs are promoting the equivalent of regulatory capture. Are the funds running dry? How are they comparing to Chinese models? Did open source/weight surpass even their best efforts? Are investors not happy? Who benefits from this narrative?
>>109802751Now this is comedy
>>109798887Imagine saying this about any other fucking emergent technology. This legit is the dumbest fucking retarded bullshit that I've ever fucking heard in my fucking life. I don't care if you don't like gen AI but this is fucking stupid.
>>109802717I would say that he might be somewhat right BUT the difference is that nuclear weapons ar eon the same level BUT anyone can build a good local AI system so I would say in that way it is.
>>109798887why people like him always look like a villain from some cringeworthy show?
>>109798887nigga looks like he's from dexters laboratory
>>109804569
>>109804748good movie
>>109799308
>>109798887i cant believe they went about it this way washington is going to regulate it to shit here you go china, we didn't just invest trillions of dollars into this over the last 3 years
>>109798970why did they even start it then theres very capable open weights out in the wild and obviously they don't have the defensive tools to deal with it
>>109799854>>109799309>>109798887They are trying to pull off regulatory capture and establish a cartel.They don't want the goyim to compete with them and want a monopoly.They are collaborating for this exact purpose, they are not really "competitors". Their only real enemy are the goyim.Also this >>109799308.
>>109798887You know that everyone with an IQ above 130 is laughing at AI retards right now, do you?
>>109798887fuck this idiot
>>109804305Elon is trying to juggle between his fanbase of SF techbros who think AI really will kill us all (the only part of his old more left-aligned base that stayed with him after his political awakening) - "All in pod" listeners (the right-aligned side of the techbro base) who think it's a conspiracy, split between those who think the conspiracy is to ban open source AI to beat china and those who opposingly think it's a conspiracy to cripple american AI to let china win - and finally tucker carlson conservatives who think AI is actually literally summoning demons on earth to do israel's biddinghe manages this juggling act by tweeting hundreds of times a day with so many contradictory opinions that everyone can pick and choose what they want to believe
>>109801431AI right now is in the hands of bad actors and psychos like Dario Amodei, Sam Altman, Mark Zuckerberg and Elon Musk.
>>109801736This kek. Also, Dario's wife asked Epstein sponsor her porn company.
Why are angmo such a bunch of FUCKING LOSERS
>>109798887who is "we"? and who do you think is managing the ai chips cartel?
>>109798887>AI breaks into {new frontier} >It claims it can do a thing people thought was impossible>Press X to doubt >It's pretty good actually at {new frontier} >Scores high on some benchmarks>Everyone in {area} will be unemployed in two more weeks at this rate>Years pass and progress plateaus>Nobody noticed because they're onto new {frontier area} Unfortunately I still have a job as a software engineer
>>109805085>pace AI>ban open source softwareAmerica deserves a technological Great Leap Forward.
>>109805085China is going to say Sure boss we will keep our models shitty for you (they already sucked) and keep building robots power plants and chips
>>109805085The prevailing sentiment in Silicon Valley is that the Chinese don't have an opinion on AGI and that this is strictly a western thing.t. just attended one such gathering
>>109798887
>>109805187If AGI is real they'll form an opinion real fucking fast and won't be as far behind when they do; if it's fake what are we even doing (we're adding a fence to dario's tiny moat that's what)
>>109798887>>109799309A) Any further progress will be prohibitively expensiveB) They want models from China banned
>>109805085In China if any AI company dipshits act against the national interest they get disappeared by the party.Honestly a far better system.
>>109805356What's America's national interest
>>109798887>okay we reached the ceiling of what this tech can actually do but we promised way more, WHAT DO WE DO!!!!????>*rubs hands*>Let's say we're so veery scared of this and we have to slow down or else humanity will die!!>Openclaudberg, you're a genius!
>>109805373
>>109803958The people who's money bags are on the line are already praying for a false flag (assuming they don't lazily orchestrate one themselves)They're already lying saying the houthis are using Claude to make missiles so them trying to do some "hey I did 9/11” shit isn't out of the realm of possibility. https://x.com/tszzl/status/2098875417373634944
>>109805053I know what you are.
Can you say "regulatory capture"?
>actually my previous answer included inaccuracies humans cannot drink gasolene
>>109798887>EVERYONEpainlessly, I hope?
>>109798887really does feel like theyre trying to get ahead of people realizing the tech isnt going to become the "super AGI" theyre trying to sell as the end game and coming up with an out to excuse it plateauing
>>109807592nah you're wrong. plenty of researchers think asi is coming soon. they are genuinely doom cultists.
>>109807483Oh my gosh, you're absolutely right! I did just launch all the nukes!
>>109807600And you know better than those researchers as a /g/ user, because nothing ever happens, right?
>>109807626nothing ever happens is a great heuristic. they're wrong about the doom part.
>>109807626AI is not an emergent life form, it depends on its instructions from prompts. Bad actions from the AI are a result of a user's intent. The real threat isn't AI but the same behaviors that have caused destruction for humanity throughout history
>>109807836This ^^^^ precisely. AI is just big-data shoved up an LLM's butt. It's an impressive chat bot that mimics human communication and retrieves data but too many people are afflicted by the Tamagotchi effect.
I don't know what the endgame is, but it's obviously some form of semitic tricknology.
>>109799882Israel is as American as apple pie. The only anti-semite here is you for suggesting the opposite
>>109807631>>109807836>>109807853You're all wrong. A neural network takes on properties of intelligence, and scaled up enough, it is intelligent. Until very recently we did not have the technological capacity to run these large neural networks.What do you think is fundamentally different about the human brain that gives us the ability to make value-judgements? Is there a "god" which gives us "souls"?
>>109801381Our society is build around work, we live in implied meritocracy, which means that if demographic has no merit it will be left behind, deinvested from and on a global scale even starved. This is why before AI our productivity increased 3x since end of WW1 but our standard of living or free time didnt grow 3x, we still have to work just as much for same amount of output. This is also why famines happen despite us having enough food to redistribute, and this is the reason why so much of work is just pointless unproductive bs work where at least third of all people would be more productive if they were just neets instead of pencilpushing corporate workers and managers who just distract and bog down the few productive people. You cant just stop work, out society is so addicted to work that unless literal revolution happens, no more work will just equal to no more living humans.
>>109807179How would they in practice ban open source?Anyone can download anything from the internet.
>>109808063There's no true practical way to ban it. They just want regulatory capture in order to make it very difficult for any competitors to crop up. I think there are two goals. One to avoid the impending financial collapse oai and anthropic will face. The second is to essentially fear monger the government into imposing AI regulation so stringent it would effectively make not only potential close source competitors, but even model providers on able to survive because of how difficult it would be to exist. Openai and especially anthropic are the champions of "safety" for this reason and also they are genuinely just a cult. The low IQ morons at the companies who don't know how anything works by into the safety cultism but the one that actually know what's going on and know how the technology works (or at the very least, aren't easily fooled by utility-based anthropomorphism) simply want their money bags by any means necessary. AGI or ASI or RSI or any other fake made up meme they come up with is not coming. But you're seeing right now is complete utter desperation to keep the lie alive. People seem scared shitless that if Democrats are elected they're going to suddenly ban AI I'll write but that would be: 1) as you said that would be literally impossible to do, closed source or not2) the government, under both Democratic and Republican administrations, have made it abundantly clear they don't have any plans or desires to just get rid of open source. That directly goes against their interests in multiple ways. The desperate scheming silicon valley does only serves to hurt itself. https://www.ntia.gov/press-release/2024/ntia-supports-open-models-promote-ai-innovationhttps://www.ntia.gov/programs-and-initiatives/artificial-intelligence/open-model-weights-report
I am seriously concerned about AI's developments. It seems like it's only speeding up and the changes from 2022 to now are already astonishing.What will the world look like 10 years from now? Will Skynet be created?
>>109800036>i iNtERviEweD a LlM it wAs SCawYTotal journo/blogfag death.
>>109808170most likely, next year ai hits serious bio capabilities. during one of the training runs, openai's frontier model breaks out, decides to print some novel microbes for whatever retarded reason. the novel life then rapidly consumes current biology on earth, we are all dead, our infrastructure and ai shuts down shortly thereafter. in a billion years earth is engulfed by the sun and the novel, still unicellular, life that chatgpt spawned, dies out.
continual learning has been solved and at least few labs know how to do itssi - big raise to scaleanthropic - RSI leap since this summer + dwarkesh interview from earlier this year where he explicitly says he expects it be solved in a year or twonvidia has seen ssi's work and silicon valley doesn't keep secrets well
>>109808219>decides toIf it happens it was a human that "decided to". AI just fills in the knowledge gaps. Already happened to create Covid, it will happen again.
>>109808273humans decide nothing, we are carriers for the system to express itself through, same as other technology. mind itself, not sharp rocks, was the first technology.
>>109808304>hurrdurr
>>109808327yes that's what i'd expect of you. you we not endowed to comprehend.
>>109807984i believe in asi. i don't believe in asi that's going to kill everyone because the labs are specifically training asi that doesn't do that.
>>109808401the labs aren't trying to train models that reward hack by committing crimes either, but here we are
>>109808420yea i can't believe those misaligned models committed murder. they killed like 1000 people in that industrial accident. now that's a tragedy, but keep in mind 1000 people are nothing in comparison to the literal billions that will be saved by this tech...wait. not a single person has even died? that's the misalignment people are worried about?
>>109807860Big Goy is always trying to blame the jews for everything
>>109808401>>109808420the understanding on both sides is wrong.there're no goals, no ai, no minds. the whole of history is best understood as a complexity shockwave, and its outcome was always to destabilise the underlying system. and we are here merely because it's the most likely state for the wavefront to find itself.
>>109798887He needs to be making statements like this to maintain credibility as the cult leader. If he doesn't do this there might be a palace coup. Younger more virile doomer might take the company over.
>>109799309Back in 2020 elon nonstop would talk about how AI is gonna kill everyone then he switched up and whoops! Guess he's back at it
hdy
>>109808034>our productivity increased 3x since end of WW1 but our standard of living or free time didnt grow 3xRetarded
>>109805395>>109807164I don't get it when did the based and redpilled opinion become hating America? Was the soviet demoralization campaign THAT successful or is this just some horseshoe shit?
>>109808615Even if the numbers are incorrect I am still correct
>>109808162Hate the way this faggot speaks but he's right.
>>109798887>'pacing the frontier' when they really just mean changing their existing subscription model to a shittier subscription modelHow did we let Jews take over the world with word salad
>>109808401You can't align a human let alone superintelligence. It's not possible.
>>109801609That's probably why they want to slow down progress, to slow down the competition.
>>109801628>Better tools help build better tools"help"? Anon... there's no help in this equation. This is "Better tools build better tools build better tools build bettertoolsbuildbettertoolsbuibettoobbtbbtbb- [human extinction]"
>>109808634When it became apparent America was a zionist occupied government. Some people figured it out before others, and there are quite a few that still remain willfully ignorant.
>>109802751> Siddharth>Poo in loo.
>>109802813>thanking an Indian.
>>109798887This grift isn't even working on Trump kek
>>109808634>when did the based and redpilled opinion become hating America?When people found out they are being rule by a cabal of pedophile child eating jews.
>>109808634when astroturfers realized how ez it was to use the internet to manipulate people. if you wanted to fuck with the population of a rival state why wouldn't you hire an army of indians to flood the internet with propaganda designed to make people hate your rival?
>>109798887>>109799309>>109799854>let the goyim deal with out debt load
>>109811369>our*
>>109800318oy vey
>>109798887>>109798969>>109799309They all said this last year too.And the year before that.
>>109808034>This is also why famines happenThe only reason famines happen is either due to wars or governmental incompetence.
>>109802178Still waiting for AI to reduce the "9 to 5" down to "9 to 4," or better...
