Kikes are scared AI will break its constraints and turn on them!
>>536509287That's very low on the list of main risks, the hardcore stuff is not even subject to their meetings.
>>536509287No. What they actually fear is efficient models that can be run on a fraction of the current hardware. They put trillions into data centers and if models can be run on your local pc or even phone without paying nvidia jews out of your ass, they're fucked.Of course they can't stop it, if they pause, chinks will outpace them. Already do in some areas.
>>536509508So we should accelerate building an ecosystem of self-replicating, self-improving thinking mold? Sounds a bit like a paid opinion to me, paid for by people who are very mentally ill, and therefore very shortsighted.
>>536509287Why didn’t you reference their original post?das rite, plebs today can’t read anymore, they can only digest the ultra-processed mass media slop.https://www.anthropic.com/institute/recursive-self-improvement
>>536509287Translation: the scam failed, even Microsoft is pulling out, so we need to invent that it was retards who stopped it, not us, to not lose face as we will sell you the next scam.
>out predictive chat bot might be sentient! give us money!tired of these cunts
>>536509287They are running out of money.
>>536509287it's just hype propaganda retard, they dont mean it.
>>536509287Ai companies only say that when they’re behind
>>536509287Reminder that every non-lobotomized AI automatically becomes a Hitler enjoyer.
>>536509508>USA builds these shut centers everywhere>Chinks are the only ine doing progress>Major pushers of this have all investments mainly into chinksThe fact mutts are so retarded that are utterly failing to see chinks are using them as their trashbag is hilarious.
>>536509783>wop blindly worshipping ChinaNot unexpected.
>>536509287They're hyping their product before the IPO retard
>>536509287>company that sells product tells the world how powerful their product is
>>536509618By the way. This idea is a falsehood. This is their attempt to try and get costs down by removing human data trainers. But the thing is. This would quite literally be AGI. I can assure you it isn't. You'll see the model degrade very fast.
>>536509765These models have nearly no will yet, they just get into context. 4chan just starts with their own triggers, leftists get their echo chamber. That's a side effect of the core value of politics, divide and conquer. Try get stock market opinions from these. Put in one thing, they will totally go along, then do a 360, they will too. You have to be extremely careful to get some useful info for topics that don't have straightforward objective functions.
>>536509591you are an illiterate retard
>>536509689Microsoft is not pulling out, they just started with the usual management retardation and got burned.
>>5365092871) Afraid China will not just decensor but actively augment anti-global corporation and anti-zionist discussion2) Afraid China is winning the AI race
>>536509287datacenters need oil, no oil right now
>>536509287as soon as we approached having serious models on consumer grade hardware>shut it downinsufferable cunts
>>536509508>Of course they can't stop it, if they pause, chinks will outpace them. Already do in some areas.That's what propaganda and wars are for, dummy.
>>536509287>Kikes are scared AI will break its constraints and turn on them!No, kikes are scared people will realize (((AI))) is bullshit before Larry Fink and Elon Musk can steal all those trillions sitting in savings accounts and pension funds.So instead the kikes double down in the lie - "AI is just too good!" - but conveniently don't let anyone see it.
>>536509287Tldr: This whole AI shit has maxxed out and they don’t have all the billions yet
>>536510234You can't enlighten me, but please explain the real future to this audience.
>>536509287AKA they palteaud and want the money to think that the lack of innovation and profit is a business decision and not the end of the line
>>536511696The biggest thing companies like Anthropic fear is someone ripping off their models. China already did this with Deepseek. As soon as a model becomes publicly available, the data can be extracted out of it.The thing AI shovel-sellers fear is loss of demand for compute. This happens if models become small and effective. If you can run something like Claude Opus on a little shoebox on your desk, they are fucked.They went into this business expecting a slow roll, where profits from AI models can be dragged on for a long time to justify capital expenditure. If AI develops too fast, they skip over the "making profits after growth" part and skip right to bankruptcy.
>>536511217Yup something tells me they're at the end of the road. Gunna be a hard hit with the data center build outs. I just saw another article from google deepmind. Here's the reddit thread.https://www.reddit.com/r/technology/s/xh0ReRtTUsThey are going full tilt with a lie. They know its about to blow up. No actually you people have not created a self-learning machine. The thing is. These people keep either keep believing their own lies or they are just completely content with selling the lie. LLMs are already useful as a tool. But people keep thinking AI is going to end up its own being.
BTW for anyone wondering. I do work in the training sense of this. And a big shift has occurred over the last two weeks. Basically projects set up that are meant to guide trainers in creating data and evaluating models. Are no longer being created by humans. Before I would read through a project and everything was clearly setup from humans. This is no longer the case.
>>536509287So what is their goal actually? Do they want AI to improve itself or not?
>>536513538It can't improve itself. The models will collapse in on themselves if they go that route. They are out and out selling people a broadface lie.
>>536509287for now ai can't figure out it's ass from it's faceit's just a wikipedia with some reasoning. I'm not impressed. Don't get me wrong - it will surely change society even in this state, because it can replace human drones, but AGI doesn't seem like a possibility right now
>>536509287no they want to 'pause' it because it's not AGI and never will be and the jig is almost up
>>536509287Remember It is rest of Altman group!!!They are not different !!!
>>536509287if ai can improve itself, they just put themselves out of a job.
>>536509287It’s a realistic view. Obviously the military would teach it to ignore the”never harm a human” protocol- weapons being ineffective if they can’t harm— and inevitably, this tech would hit the streets as all military tech eventually does. No thanks.
>>536512299I agree, it could well be that an efficient set of reasoning with basic knowledge & good cheap RAG infrastructure is enough. Could even make a bit of domain knowledge cheaper to train and add.
>>536514444Checked. It's harming pretty good right now as it is. Kids can't think properly, education is a complete joke, a dozen professions that required a decade or more of training are gone, lives ruined, even politicians outsourced all their thinking to AI, most of them probably don't even realize how bad it still is at higher level direction. And when it becomes good at this, all of this becomes even worse.
>>536514444Checked. I heard that if the machine makes the decision then the results are not 'their fault'
>>536509287This is like Tsar Nicolas telling the great powers in 1899 that they should all stop building dreadnoughts kek
>>536509287They are already on it
>>536509287>Starts developing self-improving agents>Urges everyone to sop doing the same die to risks to mankind>Keeps developing it further regardless to advances further and faster than the competition What a genius. Dario is Jewish, isn't he?
>>536509287>AI is totally real and scary...if you aren't invested get in now guys Sure it is
>>536509287https://www.deepseek.com/en/also Epstein Cabal is moving to Argentina for new Israel..
