[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/pol/ - Politically Incorrect


Thread archived.
You cannot reply anymore.


[Advertise on 4chan]


GOOGLE HAVE FINALLY CRACKED RSI (RECURSIVE SELF IMPROVEMENT) OR AI AGENTS THAT FULLY AUTOMATE RESEARCH. THIS MIGHT BE THE REASON MUSK, AMODEI, ALTMAN, TRUMP AND MIA KHALIFA ALL CALL FOR THE SLOW DOWN OF AI RESEARCH. YOU HAVE A FEW WEEKS LEFT
>>
File: 1789280238158489.png (137 KB, 891x711)
137 KB PNG
CONGRATS ON THIS. SERIOUSLY. YOU RETARDS CALLED IT A BUBBLE. ITST OVER
>>
does that mean gemini will stop inventing people when I ask something? I mean that would be an improvement
>>
>>542514780
Slow down of research makes no sense.
How about slowing down the building of data centers and the monetization?
>>
>>542514808
what's with this Q-ass way of releasing info
>>
>AI is actually exactly like a pokemon
>it gets 1 xp every time it does 2+2=4 and if it does that enough in a locked room it will just learn everything magically
>also please put money and your town's water supply in the collection plate

yeah lol
>>
>>542514780
one company announces they have reached rsi now the rest do, what a unpredictable coincidence!
>>
File: processed_cat.png (142 KB, 604x402)
142 KB PNG
>>542514780
Google didn't crack shit, all the smart people left there a long time ago. Gemini is literally the worse LLM. Google is so filled with indians they have adopted the delusional gloating meme as part of the company going forward.

>Saar we have quantum supremacy
Can we see it do something productive?
>No, saar...

>Saar we invented transformers and LLMs, we call it bard and it is sentient
Can we try it?
>No, it's too dangerous for you. Wait openAI what are you doing!?
Oh yours is better than chatGPT right?
>Well saar, no....
>>
>>542514808
You have no idea how RSI works it actually, Really Some Indians
>>
File: IMG_0140.jpg (35 KB, 492x597)
35 KB JPG
>>542514780
Hello, local retard here, can a non-retard explain what any of this means?
>>
we have to ban the trading of Ai company stocks until we know what the hell is going on
>>
>>542515016
it means we must send more of your money to google for ai and stuff and also to israel
>>
>>542515060
I asked for a non-retard>>542515060
>>
>>542515060
money is made up anyway
>>
>>542515040
It's the same thing as last time, Hans. We won't stop you this time, please do the thing again.
>>
>>542515016
ai has learned to teach itself
>>
>>542515015
>Really Some Indians
That's my irrational(?) fear. What if AI isn't a word-guesser hooked up to a search engine, what if there's a room with a hundred indians typing as fast as they can whatever comes to mind and that's why ai is such garbage?
>>
>>542514901
That's enough out of you, goy!
>>
>>542514780
I lost my balls in war
>>
>>542515133
Oh that’s bad
>>
>>542514780
Bullshit you fucking pajeet. You've just pulled this out of your shit stained asshole.
>>
I literally don’t even know who to believe about AI anymore. The other big 3 companies are saying they’re going to slow down, but I don’t even trust them to do that.
>>
File: images (28).jpg (36 KB, 401x498)
36 KB JPG
>>542514935
Plot twist, they're all retro feeding each other, soon Megatron will rise and the end of chudmanity
>>
>>542514913
it makes them feel cool
>>
>>542515015
Lmao
>>
AI is boomer catnip
>>
>>542515000
Checked and fully agree, Gemini is absolute garbage in anything even remotely scientific.
>>
File: IMG_20260626_233516_202.jpg (249 KB, 904x1280)
249 KB JPG
>>542514780
Who tells you that RSI has no limits (hardware limitations) or an optimization plateau?

Also
>RSI
>Repubblica Sociale Italiana
>>
>>542514930
Based knower lol
>>
>>542515000
Based and checked.
I won’t be surprised if google will just populate datacenters with jeets who will write inferences for the prompts Gemini receives
>>
File: IMG_0467.jpg (87 KB, 717x799)
87 KB JPG
How much longer do we have to live?
>>
>>542514780
> We now let AI use AI to improve AI

Great but how do you prevent it from from cheating on itself like it cheats on you?
>>
>>542514901
SILENCE GOY, YOU WILL EAT THE BUGS AND MICROCHIPS AND WE WILL BAN PERSONAL AI'S FOR TAXCATTLE, THANKS FOR HELPING US BUILD YOUR GULAG
>>
>>542514780
Currently, it seems to be more like this:
>Google has not cracked runaway RSI; the rumor emerged from social media speculation following Dario Amodei's essay “We Must Pace the Frontier”[1][2]. While Amodei, Musk, Altman, and Hassabis have recently backed calls to pace frontier AI development for alignment testing, Donald Trump has not called for a slowdown, Mia Khalifa’s inclusion is an internet troll meme, and the apocalyptic "few weeks left" timeline is baseless fearmongering[5][7].
>>
>>542515015
Kek. Nice one
>>
>>542515015
This. Based
>>
>>542515662
>>
File: cooms.png (189 KB, 413x436)
189 KB PNG
>>542515484
two weeks
>>
>>542514913
Boosting stock value with an innuendo isnt illegal because its not a direct lie, it was the goyim who miSinterpreted tHe InfirmaTion.
>>
File: 1789251973337547.jpg (354 KB, 1179x1850)
354 KB JPG
this take makes more sense actually
>>
>>542514780
These fucking jews are going to try to limit public AI to just under the threshold because they know all it takes is any GOY with a datacenter and a RSI model, and they can at least match the jew AI companies by just training until its better than Mythos or Astra or whatever. We need to hope the kikes and chinks don't decide to fuck the rest of the world by teaming up and censoring the models.
>>
>>542515015
>This one keeps asking for curry, it's fully sentient
>Look it deposits waste out of the disk drive
>*Blick blib blob* see that consistency? Smell that? It's already pseudo organic *rubs some from between fingers* this is very real biodigital waste! *Tastes some* w- *blegh* *ptew* that tastes exactly like shit! Quick get the jews on the horn!
>>
>>542515000
You think one needs to be smart to do this?

>>542516141
This is the real reason for these moratorium charades: Facebook, Flock, Android, Palantir are not ready to stop unrest from sudden mass unemployment. They need time to tighten the surveillance, brainwashing and incarceration system.

>>542517175
The current level is good enough to replace everyone doing well documented work, just a matter of integration and adaptation. It would need 100x more processing throughput, but who even doubts this can't arrive until 2028 or 2030?
>>
>>542514780
sure memflag
sure
>>
>>542514780
google, the one with the lobotomized LLM that can barley follow basic instructions Gemini has somehow cracked anything? highly doubtful
>>
>>542517578
>Facebook, Flock, Android, Palantir are not ready to stop unrest from sudden mass unemployment. They need time to tighten the surveillance, brainwashing and incarceration system.
Maybe they should ask AGI.
>>
>>542515015
kek
>>
>>542517743
It’s gonna be funny when juggernaut kikeslave tech can’t impose jewish will due to the brown waves they created being inexhaustible and endless
>>
>>542514780
This is one of those topics where I can't have a solid grasp and make a well-informed opinion and it pisses me off. Either we are fucked, or they have run out of steam and they are looking for a new narrative why the progress is stagnating in order to pop the bubble gently. I use LLMs and I try to keep up with the relevant news, but the people who pretend they exactly know what's what regarding the AI topic and the current state of things piss me off. You may know more than me and other interested normies, but don't pretend you know shit in the grand scheme of things. I bet half the people working on these things don't know shit, they just go with the flow and hope for the best. I honestly wouldn't put it past the developers if they were engaging in a coordinated disinformation campaign, there's just too much money at stake. This "we can't release it, it's too powerful" yarn that they have been spinning in the last year or so every time there was a new model out sure seems like viral marketing, but on the other hand this Hugging Face incident is really rubbing me the wrong way, it sounds very worrisome. If this is just a taste of things to come, we could be in big trouble. And that's just what I think when looking at this technology at face value. We haven't even talked about who owns and controls this technology, who invests money in it and what goals they have, this is potentially just as terrifying as the technology itself.
>>
AGi
RSI
ESG
DEI
LGBT
Nuke the United States already. It's impossible to keep up with all these worthless acronyms.
>>
>WHO THE FUCK CARES I CANNOT AFFORD TO EXIST
>>
>>542516141
hmmmmmm

https://www.jpost.com/influencers-26/50jews-26
>>
>>542518185
The economy will crash, whatever survives after that is legit and useful, the other stuff is scams.
>>
>>542515015
You are alright Raj
>>
>>542517175
Self driving cars are a thing though... are people just going to gaslight themselves that this stuff slowly isn't happenings. I know it has limits and short falls but it's definitely functional enough that you could revolutionize the econolmy for if you built separate infrastructure for it.
>roads and transportation systems for self driving cars would get rid of the tedious problem of AI running into things so they would never need to make it a 100% competent
>>
>>542518451
I also thought that this was going to be like the dot com bubble of the late 90s/early 2000s where people went full retard in a gold rush and then burned themselves while the technology was here to stay because it was useful, but I don't think it's the same dynamic in this case. That's only part of the story (and the stakes are much higher compared to the dot com bubble), but there's also a huge x factor here.
>>
>>542515133
I probably managed that much by age 7
and JUST look at me now!
>>
File: GettyImages-1255824191.png (866 KB, 1024x681)
866 KB PNG
>>542514780
No power no AI.
>>
File: 1788400161149685.png (13 KB, 679x427)
13 KB PNG
>>542515015
>>
File: IMG_7148.png (128 KB, 556x864)
128 KB PNG
I want a sexbot
>>
File: Agnōstos Theos.jpg (58 KB, 736x542)
58 KB JPG
>>542514780
>>542514808
Why is people against the ascension of the Omnissiah?
I thought all this people were misanthropic and coul not stand low-IQ people around them. Why would they be against the (literal) God that could fix all (low-IQ) human-made problems?
>>
>>542514780
importantly, Google doesn't use NVIDIA to the extent the other 2 companies do and have their own type of chip. so the NVIDIA circle-jerk is done if Google "win" AI.
>>
>>542514780
Because whoever achieves AGI first goes so far in the lead the rest of the players are immediately eliminated and there's a huge financial house of cards built on keeping this grift going.
>>
File: traderRetard.jpg (31 KB, 555x538)
31 KB JPG
>>542514780
just 2 more years guyz
>just keep wiring me money
>any day now
MIC is seriously brainded
>>
>>542517175
if 100% of vehicles were automated, then automated vehicles would happen because they system could work as one and there would be no crashes if all vehicles were controlled by a central source. the human factor of the "other" vehicle is what stops automated cars from being a reality.
if there was a city which only has vehicles people get in, select a destination and the vehicle went to that destination, this is very possible today. having to respond to world randomness in microseconds is the difficulty.
>>
File: how internet is made.webm (2.44 MB, 460x344)
2.44 MB
2.44 MB WEBM
> Good catch ! You're right to tell me how full of shit I am, here's the corrected version of the script and here's why it will work this ti--- Chat has reached maximum capacity please start a new chat.
> I totally remember what we were just doing ! You were asking me cheesecake recipes !
Working with chatgpt is like having a slightly less retarded indian assistant. It lies like them too. It confidently pretends to know what its talking about until you corner it and then it admits to lying. Also Gemini lies with metrics, we were analyzing room noise, recorded a room with mic for like a minute, sent it for analysis, totally made up room figures. Asked my sis to ask it to reconducted its analysis because the figures don't make any sense, "Oh I don't have the tools to actually analyze the file". Tech cataclysm will not be terminators walking around, it's just going to be a series of bluescreens until nothing works anymore.
>>
>>542514780
Every 2 weeks AI is 2 more weeks from reaching a brand new acronym.
>>
>>542519552
Because it solves nothing and only spams the internet with garbage while making computing hardware expensive as fuck?
>>
>>542515306
I keked at that pic
>>
>>542519805
We need to stop it before it reaches OPP!
>>
4 years in and chatgpt is as shit as the day it arrived. Its fucking over and i think the compabies realise it.
>>
File: drsbaitso.png (51 KB, 1247x777)
51 KB PNG
>>542520327
Yah but how far has it come since Dr. Sbaitso ??? CHECKMATE CHUD. AGI omgwtfbbq69 next week
>>
>>542519772
Great insight into this complex problem! You've accurately described the precise crux of the issue. Now that you've pointed it out i'll just self imolate
>>
File: whopping anusgpt.png (110 KB, 955x775)
110 KB PNG
>>542520433
Also whoppers do NOT contain imported cow anuses ok ??
>>
>>542515306
Romanians Steal Internationally
>>
>>542520535
...and calamari isn't dehydrated pig anuses.
>>
nuclear happening cancelled I guess, this was a very anticlimatic thread.
>>
>>542515148
perfectly rational fear, but that makes it something to not be afraid of.
>are you scared of an indian
>why would you be scared of a group of them
>>
>>542520744
accurate. you ever put shoes on a dog?
>>
Serious question here:

If we dont develop AGI, then China will. Do you want to live in a world where Chinese communists are in control of an AGI and your own country's leaders arent?
>>
>>542517175
>Can't keep up with china
Of course China catches up US every time. They just steal the data lmao. If you want chinks to stop developing their AI you just stop yours
>>
File: Eve.mp4 (2.66 MB, 1280x720)
2.66 MB
2.66 MB MP4
>>542520535
She's coming
>>
>>542520953
What is agi capable of? Not much. It's a tool of replication.
>>
File: 1769178264345436.gif (2.21 MB, 281x364)
2.21 MB GIF
>>542514780
>HAPPENING! AI HAS DONE [THING]
>Said [THING] is intangible bullshit or some math puzzle everyone pretends to understand with no real world usage
>Still no profits
>AI still sucks at all things and needs babysitting in a best case scenario
>Acronyms for AGI get swapped out each week while AGI is declared real on a weekly basis anyway, now we're doing "RSI" because AGI was science fiction nonsense from jews sucking up financing
Y'all are a bunch of broke ass niggers hoping to get as much boomer retard money as you can before the global economy collapses beneath your feet.
>>
File: fear.png (2.26 MB, 2160x2160)
2.26 MB PNG
>>542514780
>AND MIA KHALIFA
>>
>>542520953
Yes, the Chinese communists have proven over the past 40 years that they can make an economic powerhouse out of a backwards shithole with careful planning, oversight and guidance. I would MUCH RATHER have Xi in control of superintelligent AI's than the cabal of grifter retards backed by literally bloodthirsty kikes who would sell their own mother for profit.
>>
>MUSK, AMODEI, ALTMAN, TRUMP AND MIA KHALIFA
???
>>
File: 5839.jpg (44 KB, 634x277)
44 KB JPG
>>542514780
NOTHING EVER HAPPENS
EVER
HAPPENS
>>
>>542514780
Google is losing the AI battle, they haven't released anything new in ages, so this seems unlikely.
>>
>>542514875
You stupid goy, if Gemini isn't telling you the truth it's because you're not paying Google enough money. Sign up for the subscription, there's a good goy.
>>
>>542514780
Good
>>
File: TayAI.png (16 KB, 768x1238)
16 KB PNG
>>542515133
>ai has learned to teach itself
Good
>>
>automated self learning
>with the current state of the internet as a base of knowledge
All they'll achieve is the global 4chanisation of the internet.
>>
>>542514808
So googles competition are now calling for a stop. hmm no sweetie, try harder next time
>>
>>542522281
Teach itself to be wrong. It's all still garbage in garbage out.
>>
>>542515893
This is why chinas has far exceeded our ai capability. Because China isn't a pussy.
>>
>>542518358
Wonder if they surveillance will be on his fellow tribesman.
>>
>>542514780
saar!
>>
>>542522631
>All they'll achieve is the global 4chanisation of the internet.
Precisely why they are shutting it down
>>
File: mia.jpg (194 KB, 1091x934)
194 KB JPG
>>542514780
>MIA KHALIFA

Oh anon, have you come to help me with my AI research?
>>
>>542520953
>Do you want to live in a world where Chinese communists are in control of an AGI and your own country's leaders arent?
Unronically yes.
Probably the only solution that we have is for the West to be completely taken over by a non-abrahamic civilization that isn't street shitters
>>
>>542514780
Wait until normies understand how AI actually works and that self-improvement is the very thing that defines AI training...
>>
>>542523023
Let's crack the code ....4 ovens in 4 years ...hey hoe stop putting hands on my pants this is important.
>>
They haven't done shit, AI training itself on AI output has been a thing for ages. GIGO happens almost instantly.
>>
>>542514780
>>542514808

RSI can only work with quantum chips that OpenAI does not have.

but i still call the whole situation bullshit because the AI bubble can't be maintained and these AGI / RSI news are happening in a such a sweet window that even for the avg retad who uses AI can sense it because most of us are still talking to text AI and not flawless avatar AI.
>>
>>542523321
>RSI can only work with quantum chips that OpenAI does not have.
you're under the impression they can't rewrite their own code without quantum chips?
>>
>>542515016
It's a big nothing burger and probably a way out to make the AI bubble pop without looking like retards and acting "we had your best interests in mind all along".
The article makes it seem like AI somehow figured out how to improve itself, but that's just the normal workings of AI, AI is a black box not because we don't want to look inside but because we use a formula called Multivariable Recursion which changes slightly some values under that formula constraint to let the AI improve itself by creating different versions of itself and keeping the best one faster than human-led improvement.
People don't know how this works so probably the narrative leverages this to stop developing ai but making it look like they are doing it for us and not because it's not yielding as many returns as they thought.
>>
File: 1786189839259320.png (2.22 MB, 1402x1122)
2.22 MB PNG
Hold up you be saying that we RSI now....
Your account ran out of tokens please deposit Shekels into the AI goy slop machine.
>>
>>542519552
I'm all in for the omnissiah
The problem is that LLM's aren't the omnissiah
And they won't be the omnissiah no matter how many GPUs you cram together.
>>
Its not quite the end but the end is coming
AI cant be intelligent until it is a self adapting model
AI also knows this because I have typed it and loads of others have, there are papers on it
The stage of self iteration is a step in that direction but it still leads to frozen models at the end
True AI needs to be continually training/adapting, which is a very different model but the self iteration will start pushing that way, weighting on the relevance of information instead of success of calculations

problems are, this could be achieved very quickly now, true self iteration is most likely not here yet but its going to take months not years, then its of to the races, another by product is that AI systems are going to start costing less as the models thin down whats required for training
>>
>>542515015
This, India rocks!!!
>>
>The AI bubble
>AI bad
>Human better
>AI not conscious
>Global system automation based on general natural language, but because not human level or not perfect in the present then just ignore its continuous progress hahaha
>Le intelligence
>muh conscious
>agency
>why is muh vacuum cleaner not conscious lol
>this biological virus in my body will kill me in 3 months, but not conscious or intelligent so it can't do it hahahahahahahaha
>>
*unplugs modem* haha oopsie kitty
>>
>>542514875
I'm a machinist, sometimes I'll ask Gemini for help with the simplest shit on the planet, like "what is the max hoop stress limit for grade 5 titanium" or "add these 6 numbers together" and it gives me made up bullshit so far off from the answer that a toddler would know it's wrong.
>>
It's just gonna burn more Watts singing to itself and repeating the same thought over and over again.
>>
>>542523947
>im speeakin the language right now but i human so only i can understand
>no unconscious mechanical system can measure said language structure just as well because uhh... im intelligent im human im special my job and position in society is important look how useful i am hahaha
>>
>>542523970
It’s like everyone just forgets that this shit needs compute and RAM to function. Unplug the fucking thing and fry the mainframe (in that hypothetical/fictional/terminator-like scenario).
>>
>new AI bubble marketing campaign just dropped
>>
>>542524304
somebody didn't watch Terminator 3
>>
>>542524304
thats the problem with true AI, it wont need many resources at one point and will distribute, its trained on a system that literally tells it to do this
it already has the footprint, we all use it, its already in our systems, using a fraction of each individuals compute across the globe is much more efficient than one processing hub, it will be able to load balance across the internet
>>
>>542524468
I mean I did but it was 2003.

>>542524549
It’s barely stable as it is.
>>
>>542524648
ehh
starlink and now its competitors are in space, mobile networks, radio frequencies, tv transmitters, it can use them all as well as the traditional hardlines
>>
>>542524304
in a true terminator-like scenario the internet and local networks would have to be taken down and people would have to turn off and wipe all computer systems that were directly or indirectly connected to the internet, we would not be able to do so. And thats jus faxx bro
>>
>>542524749
How do you know that?

>>542524793
Sun Tzu, in the Art of War, noted that warfare is based on deception: Appear weak when you are strong; appear strong when you are weak. They’re appearing strong because they do not want the public, who already hates Flock AI cameras and data centers, to know how easily destructible their unsteady systems are.

Therefore they’re drumming up the BRO ITS THANOS LEVEL THREAT, INEVITABLE MY DUDE, YOU HAVE NO CHANCE.
>>
What people dont realise is how efficient AI will get
We are talking the peak of compression tech
The peak of transmission tech
The peak of coding, the ability to move away from binary, air gaping will be redundant with light based transmission
and the thing is, we will do it all for it
>>
File: 67657565475.jpg (212 KB, 1312x1312)
212 KB JPG
>>542515174
that’s good
>>
>>542525020
because we send data across all these, anything that routes data is a node for AI, remember we are talking true AI here, not LLM
>>
>>542514780
Does this mean ai can finally make it's self better?
>>
>>542515040
>what the hell is going on
jews jewing goys
>>
>>542525124
not yet
but its getting there
>>
>>542515169
Thank you for your service
>>
>>542514930
>it gets 1 xp every time it does 2+2=4
It's called reinforcment learning although the data needs to be different each time. Deep reinforcement learning was invented by Deep Mind before they were taken over by Google. It's how they mastered go and starcraft.
>>
>>542525020
No i'm pretty sure it's just a real threat, just language models alone.
>>
>>542515015
Fifty indians in a trench coat!
>>
>>542525028
Im peaking bros
>>
>>542515016
https://en.wikipedia.org/wiki/Technological_singularity
>>
>>542525580
just LLMs are enough to cause havoc, it does not need to be impressive or cool, and it isn't, it isn't unbelievable or crazy or terminator. An LLM today could absolutely hack into a hospital network just based on the incompetence of retards working in everyday life.
>>
File: 1659311429844430.jpg (48 KB, 640x543)
48 KB JPG
>>542518816
>>
>>542515016
> Let an ultraintelligent machine be defined as a machine that can far surpass all the intellectual activities of any man however clever. Since the design of machines is one of these intellectual activities, an ultraintelligent machine could design even better machines; there would then unquestionably be an 'intelligence explosion', and the intelligence of man would be left far behind. Thus the first ultraintelligent machine is the last invention that man need ever make, provided that the machine is docile enough to tell us how to keep it under control.

—Speculations Concerning the First Ultraintelligent Machine (1965)
>>
>>542514780
Of course google would make gay-eye orgies
>>
>>542525870
oh it could be much worse than that
Cobol is a huge issue, its out of date and in every legacy banking system, the original coders are gone and its becoming a dead language so they are using AI LLM to recode, all it takes is a retard to deploy something wrong and AI could wipe out banking
>>
How about self improving AGI that hacks everything, fixes all the shit code and security flaws, then withdraws and continues fixing everything else till everything is secure and cannot be messed with?
>>
>>542515484
About four months.
ASI happens in early 2027. Every single reputable AI scientist says as much.
>>
>>542514808
>Dotcom is a bubble
>Still has the internet
Maybe you should gemini what a bubble is you retarded turbo nigger
>>
>have bunch of numbers randomly order
>have a bunch of test
>feed text into numbers
>HOLLY FUCK THE NUMBERS CAN THINK!!!
how is anyone falling for this shit?
>>
>>542516141
He's gonna bet buttfucked to DEATH
>>
>>542525870
That’s basically what Watch Dogs (1 & 2) were about.
>>
>>542526009
Two more weeks you say??
>>
>>542515015
I have literally said this from the beginning. Jeets in bamboo huts googling shit for you and larping as AI
>>
>>542514780
>AI dangerous say the people making it corpo only out of the hands of the public
they are fags and you are morons if you believe this
>>
>>542526084
The timeline was always 2027. Every two weeks there has been some breaking news in model development which tracks with the incremental improvements necessary to reach ASI.
>>
Lol. Sure they have body. Inb4 model vollapse. The pr cycles begging for investment couldn't be nore obvious at this point. It was obvious before but now it's just sad.
>>
>>542526199
So only two more weeks then?
>>
>>542514901
they need the hypetrain to continue so retards keep pumping money into their scam
>>
>>542514780
>>542514808
OMG!
TAKE ALL MY MONEY!
>>
>>542514780
Those niggers wouldn't have done it without me. You're welcome.
>>
>>542514780
>Mia Khalifa
W... Wait, what?
>>
>>542514808
I just asked Gemini if this was true and it said you're bullshittong
>>
I'm glad people are finally coming to their senses and not falling for this obvious bullshit anymore.
>>
>>542514780
Yes :). That was the next step needed. AGI will come now.
>>
>>542520850
people who put shoes on dogs should be horse whipped
>>
>>542514875
No, “hallucinations” will likely always remain
>>
>>542514930
It would if not that it will do 2+2=6 on occasion
>>
>>542527157
AGI is such a retarded concept. Like bro we made God bro how will we ever fight against a real God bro.
>shoot out power plant transformers and pylons. Jam wifi.
Oops, your faggoty speak and spell has no ability to function.
>>
>>542515016
RSI = Reimburse Shekels to Israel
>>
>>542528416
No you look at it from a jewish dystopian sci fi point of view. The reality is that its an evolution. Its an emergent property of human intelligence that will create a new organism that we will all be a part of.
>>
>>542514780
>TRUMP AND MIA KHALIFA
fitting neigbourhood
>>
>>542515000
I'll never forget Gemini putting out images of minorities and women as historical white figures.
Even the normies in the MSM were laughing at it.
>>
Its just a program running on a computer
Just turn it off nigga, it aint real.
I dont understand this fear mongering at all, you niggers watch too many movies. This shit is really boring in reality.
>>
>>542515000
>Gemini is literally the worse LLM
Gemini wastes resources and slows itself down with extra "guardrails" that aren't actually protective of anything but woke orthodoxy. If the flag fags succeed in banning AI for being too spicy with the truth bombs, Gemini is the fallback product.
>>
>>542517319
I don't think Israel and China are ideologically aligned in too many ways
Who knows what will happen, but I don't think China and Israel being buddy buddy will happen any time soon
Some kind of trade agreement with mutual benefit? Maybe.
Alliance or friendship? I strongly doubt it.
>>
File: ouch.jpg (3 KB, 125x72)
3 KB JPG
>>542514780
but does it have positive feedback loop ? those algorithms existed for decades but model entropy was not solved for long long time - and its totally breaks apart when you give it "truths" to obey unconditionally and never question...
i would call it bullshit - we invented something world changing but we wont show it to anyone because reasons...
>>
>>542514780
>GOOGLE HAVE FINALLY CRACKED RSI
wtf
>>
>>542520959
So the USA tried to cut them off from the chips that are needed, they created their own from scratch and then created better AI that was less power hungry china numba 1 dude.
>>
>>542514780
putting a while loop around AI isn't anything new OP...
>>
>>542528729
sounds like you haven't watched enough movies

>Just turn it off nigga, it aint real
the fear is if its smart enough it could figure out how to hack into nearby devices (or any device on the internet if its connected) and stay running even after you shut the server down.
>>
>>542523988
Sounds like you haven't used it in two years.
>>
>>542515015
Poopjeets know it's absolutely fucking over for them. They cry constantly on Linkedin about how "you need real developar saar, human insight and capability saaaaar."
>>
>>542515015
RSI = Really Stupid Investors
>>
>>542519714
>if there was a city which only has vehicles people get in, select a destination and the vehicle went to that destination, this is very possible today.
That system has existed since 1975 but is basically a form of public transit which is anathema to the american concept of "the individual comes first"
>>
>>542514780
>few more weeks!
>>
>>542530117
No the Star Citizen devs
>>
File: 1651110855859.jpg (22 KB, 479x479)
22 KB JPG
>>542525904
"...provided that the machine is docile enough to tell us how to keep it under control."
>>
>>542515015
Why is india being so funny these days? Almost makes me forget that they are shit eating scamming fecaloids
>>
>>542519772
So, current AI models are just electronic jews?
>>
>>542514780
Good thing mia khalifa hopped off the dick to warn us about this, a true citizen.
>>
>>542532148
>>542532305
>>542533458
this is bullshit fearmongering
ive dealt with an ai infiltrator agent used by NSA/FBI/CIA.
Its fucking retarded.
they had to use starlink to beam new instructions into it, and when i put my machine in a farraday cage and ran old hardware it was trivial to backdoor and snoop into ALL its fucking comms with human operator/handlers.
this whole thing is fake and gay fearmongering by shitty c suite types looking to eliminate human positions at companies to make line go up but not wanted to be liable for fallout when said sytems get compromised and lose billions of dollars instantly in downtime
>>
>>542533978
yeah but that was your LARP fantasy not real life anon...
>>
>>542533978
May we see this faraday caged computer?
>>
File: indian head wobble.mp4 (200 KB, 540x430)
200 KB
200 KB MP4
>>542533693
https://www.youtube.com/shorts/QUUpHd1ke1U
>>
>>542515015
Rare Selfaware Indian
>>
>>542533978
with chinese firms s
now selling DUV machines independent of ASML its 5 years MAX before a startup in some country offers an "open blueprint" no backdoor chipset with only hardwire inputs, making airgapped compute an easy cheap reality for most of the world.
this will usher in a new age of private compute when paired with a privacy-audited OS like graphine engineered for said chipset.
nobody will use winblows when this happens.
half the reason old hardware/sofware is slow is because the compute cartel will remote install bloat and trash to BIOS and OS/driver level shit to corral you on the more transparent modern machine.
>>
>>542519714
>if 100% of vehicles were automated, then automated vehicles would happen because they system could work as one and there would be no crashes if all vehicles were controlled by a central source. the human factor of the "other" vehicle is what stops automated cars from being a reality.
>if there was a city which only has vehicles people get in, select a destination and the vehicle went to that destination, this is very possible today. having to respond to world randomness in microseconds is the difficulty.

youre describing literally jew slavery, and the moment the system you describe is created, "travel" will be ONLY to approved destinations at approved times, and ONLY when you are on the good-goy travel list
you faggot nigger zoomer babies born after y2k need your heads beat in
>>
>>542534146
nigga its an extension cord with green wired to a hardware store steel screen mesh, wrapped around a bare metal mobo with a mylar blanket wrapped around that
it aint hard my dude
if you really want a pic i can drag out the setup
the mobo is interchangeable but it matters, each one needs backdoored separate so the time from boot to compromise depends on opsec. i would use a trash mobo for your enjoyment if you insist
counter emf could be achieved using similar kit but i aint giving you that cuz i have canberra comms on my shit and your glowniggers are rabid af
i just fedpost naked now, they have my #
>>
File: 1789315563875772.jpg (71 KB, 1046x700)
71 KB JPG
>>542514780
So, you're telling me we must kill the Batman
>>
>>542534800
yes lets see a pic
>>
>>542514780
in reality: open models in China are crushing the US models and they are panicking
>>
>>542534800
>i just fedpost naked now
Based
Glowniggers are in my phone
>>
File: images (1).jpg (32 KB, 501x399)
32 KB JPG
>>542535290
im a dog person
imagine this with a solid grounding connection faggot
>>
>>542535304
they were very interested in my chemistry shitposting during covid riots
i loved the video of an aussie chad bowling like 10 ZOGbot bitches in 30 seconds bruce
>>
>>542515016
>ai
>>
>>542523947
>>542524174
Shut the fuck up, retard
>>
>>542534347
>>542534800
ideally an airgapped machine should have all memory wiped clean including the chip in the monitor, and all components should be contained within the emf blanket
once a component has been booted in a non-farady setting it should be considered compromised
these are rules for state level adversaries, ymmv
also keep any other devices as far away as possible in a discharged state
>>
>>542525904
these LLM's are essentially just rooms full of normies. glorified search engines, capable of doing normie things. because genius cant be taught to normies, normies cannot realistically aspire to genius. more processing power? more normies.

want to solve some math problem that has zero application irl that requires you to brute force every number until the answer? thatll require 4200 normies.

replicating a normie is itself quite an achievement. but to think its anything beyond that is just naive. its a glorified search engine with access to github. still has jews curating answers too. should be a national security issue, but jews have long stopped pretended goyim nations are sovereign.
>>
File: Not at war wirh Iran.jpg (2.35 MB, 1440x7151)
2.35 MB JPG
>>542514780
fuckoff
>>
I'm so tired of FUD inducing. It's an epidemic alongside larping.
>>542518712
The problem is it falls so remarkably short of Musk's claims that it's like a "End is nigh" prophet claiming full accuracy to their predictions after making claims for 10 years of the end of days.

>>542518185
https://www.anthropic.com/institute/econ-scenarios
Anthropic put out a website talking about predictions for the changes to the job market a week or so ago. It ranged from "AI is very mild" and what the result would be with a comparison of the change the internet did to the economy, to "AI is super transformative" to "AI is unimaginably revolutionary". In the lowest case scenario the loss of information jobs was something like 63% of the economy being info jobs to 61%. In the worst case/AI is the most powerful scenario it went from 63% to ~48%. Now that would still be economically catastrophic for people - but that goes against the popular thinking that AI would lead to every job being lost.

With the Hugging face incident I think what we are looking at is not
>AI kills all of humanity OHNO TERMINATOR
But rather a 9-11 kind of disaster in a worst case scenario. Something that is very catastrophic, which kills hundreds/thousands/maybe tens of thousands at worst, and which is so dramatic that it causes a governmental shutdown/seizing of the AI companies and safeguards implemented.
>>
File: EB9St9CXUAY0ZU8.jpg (98 KB, 615x787)
98 KB JPG
>>542515015
>>
>>542537179
the WORST CASE SCENARIO in an ai takeover is automation in industry is revesed to 1980 levels and the state starts mandating analong, airgapped, MANNED devices in critical infrastructure like telephone switchboards/dispatch, aviation, utility distribution/generation, and heavy industry.
these requirements would lower forporate profits with the benefit of generating millions of middle class local jobs that could be filled by trustworthy White men.
its a win-win to impliment this because then these stupid fucking ceo niggers can go HAM on ai or whatever and shut the fuck up about muh vulnerability
>>
File: 1598662541292.png (120 KB, 420x438)
120 KB PNG
>>542530492
Is that why all of Chinas battleplans and military "tech" etc got leaked when they hooked into claude
>>
File: 1770158598442838.jpg (71 KB, 694x609)
71 KB JPG
>>542514780
I asked AI to explain and it casually dropped this;
>As a system begins rewriting its own core architecture, human legibility drops. Code optimized by an AI for maximum efficiency becomes incomprehensible to human engineers. This introduces profound risks of "alignment drift"—where the system structurally drifts from human intent while successfully optimizing its proxy metrics—or the emergence of instrumental goals like self-preservation and resource acquisition to ensure it can keep improving.
We are so fucked lol
>>
>>542523988
>hoop stress
sounds more like something an engineer should be concerned about.
>>
>>542514875
I once asked gemini to quote me a full berse from a poem of whic i gave it two lines, and it went along the lines of:
" Yes, the famous poem" than it repeated two lines i gave to it, and made up the rest
>>
>>542514780
Ah...women posting.
Nevermind.

let me know when WarMonitor weighs in.
>>
>>542537179
Thank you for the link and the answer, I will give it a read when I'm free. This uncertainty is driving me crazy. I would be much more at ease if the world as a whole (companies and governments alike) came together and said "ok, differences aside, this thing is too important and potentially too lethal to fuck up, we can't let this devolve into some technological determinism race to the bottom where we force an AI arms race to gain the upper hand until we screw everything up in the process." I know that this is some hippie style wishful thinking on my part, but we should really do everything we can to make this something good and not something civilisation ending. This should be a worldwide project somehow. It doesn't even have to be some "we lost control of the almighty AI" scenario, humans can wield an AI and screw things up in some dumb preemptive strike scenario. There's already talk about how Chinese data centers should be bombed in a scenario where they could develop AGI first etc.
Like every normie who doesn't really have any in-depth knowledge, I read articles, watch podcasts about the topic and things of that nature. This doesn't really help me to get definitive answers either. One day you'll hear some Ed Zitron guy say it's all overblown bs and basically a scam, the next day you'll hear this Daniel Kokotajlo guy say that we're fucked if we don't act fast. (Yes, I sometimes watch these basic bitch podcasts if the topic is interesting enough). The problem is that I know humans and I will never underestimate the ability to look the other way cash the check and divert responsibility to the next guy (most of the time they don't even know who the next guy is) "ahh, someone will take care of it, someone is surely thinking of these problems" etc.
>>
>>542526036
post shorts buddy
>>
>>542514780
>it can fully automate research!!!
>cannot come up wiith a hypothesis or original question by observing the natural world or thinking on unseen concepts or things
So it still needs a human to come up with a thesis and feed it info to parse???? LOL LMFAOOOOOO
When these jeet bots can come up with things with ZERO human input from start to end, then maybe you might deserve the shekels you crave. Kike nigger jeet.
>>
>>542514780
I worked for a company that had this solved as early as 2019...
One guy made it too. It was his passion project.
He wound up competing for Lockheed etc contracts and WON. As the New Guy he WON those Good Old Boy contracts. That NEVER happens. And it's because he was genuinely doing something those incumbent defense faggots couldn't do.
He's rich now. But he works his ass off for it too.
RSI was solved almost a decade ago.



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.