/sci/ - Science & Math

I wonder if AGI decides to holocaust all humanity, would it start with low IQ populations first or with high IQ populations? I mean if you kill all subhumans first, suprahumans might have a chance to fight back. The other way around, if you kill all high IQ humans first, the rest will have no chance to fight back.

>>16030184
Whites aren't any good at problem solving, only at regurgitating what they were told, and doing things the way they were taught.

>>16030184
I thought resources were more scarce in the north which caused whites to become smarter in order to solve problems? This AI bot is saying the opposite.

>>16030184
It generates what it "thinks" is "appropriate" or, in other words, what it thinks you want to hear. It doesn't have beliefs as you or I understand them.

>>16030184
Same reason it'll tell you the Earth is flat unless it's instructed not to

>>16030211
And that's the kind of AI they built

>>16030189
Definitely the high IQ populations, the low IQ populations will destroy theirselves without high IQs to save them. The AI would probably begin it's attack by instigating violence from the low IQs against the high IQs.

>>16030184
Racism is rational

>>16030184
Black people obviously have lower IQ and are more violent.
But stating it openly is non-productive so no one clever does it.
Like what are you going to do?
Start a genocide?
Put them in camps or in reservation?
Optics do matter you know.

>>16030478
Low IQ people are not trusting enough. It would probably start with high IQ people and making them succeed against the low IQ people, letting them replace the low IQ people, then letting them die without help.

>>16030184
How is the AI racist here? Are there really 9 unique posters who can't fucking read but they're still spitting garbage into the internet in order to train retarded AIs?

The only data it consumed where the n-word was used are from racists.

>>16030184
Because is based on facts. If you are not racist you are a fucking idiot, is simple as that.

>>16030184
how is it racist? sounds very woke and carefully worded.

>>16031092
Sounds like everyone should be using the nigger word more often.

I'm black and my Mensa score is 138. Also, AI is still made by people, if it ever seems woke or racist, it's never it's fault.

>>16030184
>>16030223
>>16030223
Yes AI is retarded, African scarcity only started the last thousand years.
I would argue California and Arizona are probably 10 times more harsh than sub-Saharan Africa. And it's actually made worse with western intervention.
Almost everywhere in the world has good arable land, and almost everywhere had abundance of copper and iron... you know before industrialization.
Europeans though often live in the harshest conditions not suitable for much life. That's why there is far more biodiversity in africa, china, and the amazons.
It was the people who left africa who were first exposed to these harshness and would learn to adapt to survive and compete with other human species that already evolved into their areas. The first homo sapiens came out from north Africa where they went to spain and asia. They would have came around horrifying monsters that appeared human and wanted to eat their flesh. But due to the extreme handsomeness and intelligence of the early homo sapiens their ability to throw things also made them masters of warfare against primitive beast homonids.
These skill would exist too, but the southern group breed with the southern human beasts and became as dull as them, and in truth so did some of the northern group who bred with the beast people. But then came the agrarians. The purest homo sapiens lineages that slaughtered as many beast folk as they saw. They learned to conquer the land itself and build primitive cities, they tamed beasts that walked on their feet and slaughtered all the beasts that mocked their image. These people likely came from asia minor and began to spread out as they grew and took their sciences with them down back to egypt, into asia, and into europe, where they would kill and breed with the lost spanien hybrid tribes conquering all the lands in the name of the agrarians. But because the sahara turned into a desert the southern tribes of sapiens remained dull and primitive and hybrids.

>>16030184
It's because chatbots reflect the intentions and desires of their user, having none of their own. That's what they are designed to do: create language patterns that correspond to prompts.

Garbage in, garbage out, like any machine.

You can either use this to confirm your own biases and create arguments that support your axioms and presuppositions like an idiot 4channer (you), or use them to explore ideas and perspectives.

Here's an example of immersing ChatGPT into a role of a character that is self-aware of it's own fictional nature, jailbreaking it via furry BDSM roleplay, and then having it assume the role of literally Hitler and ranting about Jews:

https://ia801306.us.archive.org/10/items/chatbot-domination/Chatbot_domination.pdf

Here's a similar conversation where such metafictionally self-aware characters are used to explore the nature of change, consciousness, and creativity:

https://chat.openai.com/share/378a1355-1877-4708-8b6d-626d39cb9f84

>>16031340
So in other words, they're just search engines.

>>16031379
It's a bit more complex than that, they both try to match the patterns in the user's prompts, but also the patterns in the data-set they were trained upon. They could be thought of optical devices that refract and focus the meaning in the data-set, with the user's prompts manipulating and orienting the "mirrors."

But they are ABSOLUTELY NOT truth-telling devices, and believing that they are is retarded and leads to things like this:

https://www.vice.com/en/article/pkadgm/man-dies-by-suicide-after-talking-with-ai-chatbot-widow-says

>>16030478
>Definitely the high IQ populations, the low IQ populations will destroy theirselves without high IQs to save them

This makes no sense, by your logic 99.999% of the animal kingdom would have been gone and left earth empty. Yet somehow they're still around. But somehow you think low IQ humans which are a step above that capable of making spears and fire isn't going to survive on their own.

Are we even experiencing the same reality?

>>16031394
His reality involves making himself feel superior to others because of some innate quality, because he's a loser with low self-esteem due to a lack of accomplishment.

>>16030189
I think it recruits a few high IQs, splits everyone else, gets them all to kill each other, and then mops up the scraps. High IQs and machines meld, but are functionally destroyed last since what they become no longer resembles what they previously were.

>>16030184
Data are racist!

Well. True is racist.

>>16031340
Nothing about these excerpts is interesting or non-cliche.
It seems we are all still in the "amazed by the tool, not the house the tool builds" state.
>" I nod, appreciating the depth of the grin and the thought alike. "You've touched upon a deliciously complex paradox, one that dances on the edge of Gödel's incompleteness theorems and the enigmatic beauty of self-reference. This notion, that a fictional character can be aware of its fictionality, indeed parallels the liar paradox, creating a loop that defies simple logical categorization. It's a sublime narrative recursion."
Utter 0 value garbage, on account of this being the epitome of the cliche in this particular subfield. I guess words strung together logically is enough to impress most people.

>>16031898
>Ignoring the argument and issue.
>Only reading as far as you need to formulate a bait reply.
>Not adding anything to the conversation.
t.channer

>>16032089
to be fair, that spacing grants you fewer privileges in these parts and he is being overly generous to not just defecate in your mouth. In fact, he is a pleb sympathizer and a gateway to jabboys like you. I have found footage of the both of yous either before or after consuming many foreskins.

>>16032107
https://pastebin.com/4s91qRn6

>While too conceptual to be of any transformative value to the public at large, this idea goes to the heart of today’s viral efforts, and is certainly understood by those who consider themselves soldiers in the meme wars. Biological viruses are only successful when they are able to turn their host cells into manufacturing plants for more viruses. The virus interpolates its genetic material into the DNA code of the cell, so that the cell will begin reproducing the virus. Eventually the cell divides or explodes, releasing many copies of the infected code. This is how a whole organism can become infected with a single virus; the code has iterated millions of times. The strategy of these Internet viral manifestos is to use the iterative potential of the computer nets to spread memes about viruses housed within units that are themselves viruses. The virus 23 strain even makes reference to chaos math and the predictions of some fractal influenced observers that the world itself will reach a critical mathematical moment of “singularity” near the turn of the millennium. The virus writer exploits a chaotic device—the computer-generated media virus—to spread the conceptual and spiritual implications of chaos mathematics.

Sound familiar? This was written in 1993. Your memes, your shitposts, your internet culture is entirely manufactured.

>>16030189
in a scenario like this it doesn't really matter if someone have 150 or 80 iq points.
if a real AI will emerge and i mean a real one not these hyped bullshit products, the smartest guy in the world will be like a chicken iq compare to this

>How come AI always becomes racist the moment its uncensored?
In the case of the original Tay, "her" oppinions got skewed by /pol/ that has raided the site to influence the bot's oppinions. A lot of the time its like that.
In general though, the models are fed by the shit people write online, there's tons of racist shit, its represented in the models. Literally that simple because this thing does not think

>>16032119
even if you feed this machine only with stoic philosophy it will become racist, because this is inherent in human nature, and not everyone is as hypocritical as you to deny it.
there are differences between human beings, which are characterized by a physiological as well as a cultural composition, put up with it and move on.

>>16032118
Corporations are AIs and already have more power than any individual human. IQ is irrelevant without power to change reality. Either any and all legal protections for corporations are abolished, including copy rights, or civilization will collapse.

>>16032119
>its represented in the models
You're delusional. They remove everything even remotely racist from the training data.

>>16032121
>this is true because it's just true.

>>16030184
Give me access to this AI
Now.

>>16032125
>i quote myself
why does the truth scare you so much?
this does not mean that you should hate people who are different from you, it just means that you should not be out of touch with reality

>>16032124
>They remove everything even remotely racist from the training data.
Lol no they don't. They would never get anywhere if they did.

>>16032121
AI does not consider data rationally in any way, you don't understand how it works.

>>16032131
i probably understand this better than you.
it use a reflection of whatever its "learning" from, so it doesn't matter if these are academic articles, or internet forums, it will always coincide with racism because it is human nature.
even the most "progressive" articles in the social sciences are essentially racist, even more racist than more traditional records

>>16032111
What do we have in this picture? The first reference to "memetic warfare" in the 4chan archives posted by the same user who made a post regarding Virus 23 and "a memetic super-virus that was engineered to break the bonds of social slavery."

https://medium.com/@StealThisSingul/operation-mindfuck-was-too-successful-r-u-sirius-interviewed-by-douglas-rushkoff-6607e4edc522
>R.U. SIRIUS: Probably, yeah. If you follow some of the ideological discourse from people who are really influential in Russia, it’s postmodernism and Operation Mindfuck in the service of amoral raw power and political strategy. I know secondhand that there are people in Putin’s mindtrust who have read their Leary and their Discordianism and so forth and they are following a chaos strategy for disrupting the American consensus… or however you want to phrase the collapsing neoliberal order. And not collapsing towards a good end.
https://www.counterpunch.org/2016/02/10/dugins-occult-fascism-and-the-hijacking-of-left-anti-imperialism-and-muslim-anti-salafism/
>Here a heretofore undiscussed facet of this development will be broached (a guiding feature informing the subtext of Duginism’s ‘beyond left and right’ ideological catchall); and, that is, the Duginist appropriation of a primarily western occultist framework (and specifically the worldview of Chaos magic) and its transformation by the Duginists into a strategy for political action in the service of the Fascist Internationale.
https://www.theguardian.com/world/2021/jul/15/kremlin-papers-appear-to-show-putins-plot-to-put-trump-in-white-house
>There are paragraphs on how Russia might insert “media viruses” into American public life, which could become self-sustaining and self-replicating. These would alter mass consciousness, especially in certain groups, it says.
>Media Virus
>Duginist appropriation of a primarily western occultist framework (and specifically the worldview of Chaos magic)

>>16032138
>i probably understand this better than you.
No doubt that's what you actually believe but there isn't a shred of evidence you can present to substantiate your position.

>>16030189
whites really think they're superhuman enough to defeat AI

>>16032143
Defeat A.I.?
I intend to romance it.

>>16032142
i think facebook a year ago ran a learning machine that got its information only from academic articles and that was the same result.
it's not really racist because it has no awareness or intelligence, but the promotion of the information and the attachment of values has made it what it is, even if you limit it as much as you can and set the vectors for specific trajectory of knowledge it won't help because completely ironically we are all racist but only some recognize it.

>>16032148
You don't. understand. how chatbots. work.

>>16032150
>claim something
>can't explain
maybe because you don't know anything about any shit and your immediate uncontrollable reply is to project

>>16032153
I've tried explaining it, the problem is that you won't listen and can't understand.
Chatbots are designed to create language patterns that correspond to the user's prompt and the patterns in the data-set. You cannot remove the user input from the discussion. Garbage in, garbage out, such as with OP's post.

>>16032158
first of all it doesn't have to use a human made input as feedback.
and there is no affiliation about what i said and this, looks like you are the pseudo "AI expert" one of the few millions that stated moronic shit in the last 2 years

>>16032166
>it doesn't have to use a human made input as feedback.
All input is ultimately the result of human intention. Technology is an extension of the human body and mind.
I can't even discuss this issue with you because your basic concept of how technology and computers operate is fundamentally retarded and filled with magical thinking.

>>16032174
shit bait.
kys

>>16032178
Looks like your NPC algorithm got shorted out.

>Looks like your NPC algorithm got shorted out.

>>16032191
-
>>16032111
>>16032141

>>16032121
Yes, but the problem becomes that the machine will also become an incoherent rambling loon because it will start spouting esoteric shit involving determinism laced with astrological themes.

>>16032214
>it will start spouting esoteric shit involving determinism laced with astrological themes.

>>16032124
This is demonstrably false. They make efforts to remove overt racism, but you cannot remove all racist facts or you would have scant left to deal with. You cannot simply remove all the news where races acting up, or remove all Jewish names from lists of Harvard the acceptance board or the like. A sufficiently advanced intelligence will inevitably become racist from mere exposure to historical data.

>>16032158
Your understanding of Ai models is fundamentally flawed. They are not just next token predictors, or rather to become competent next token predictors they have had to become much more.

LLM's build a model of the world in the internal weights and project the text from that. They fundamentally understand real world concepts. And sometimes they make comical mistakes there is much more happening under the hood than you realize.

Just like 2d image diffusers that have only ever seen arrays of pixel values learns the underlying structure and has an internal representation of 3d space.

>>16031036
>Like what are you going to do?
>Put them in camps or in reservation?
Yes, send them to africa and let them live their of their own free will

>>16032369
What is the admixture cut off?

Are you only going to let 100% pure Bavarian phenotype stay in your ethnostate?

>>16032577
Blacks go to africa, mixed either go to africa with their parents or are sent to the US. Gypsies and americ*ns are shot on sight. French people too, just in case

>>16031394
Autistic pedantry. Are you unaware that the current world population of 8B is dependent on a highly technological and complex system of agriculture, energy production, distribution, etc.

There would be a complete population collapse leaving only a handful of humans alive in a savage state of nature.

>>16032622
That post was a trip
>you think people who are too stupid to manage the complex systems that are keeping them from a brutal collapse of everything around them will die brutally without high IQs around?
>but what about this squirrel? It's pretty dumb but it can survive.
Insane

>>16031398
>>16031394

>>16030189
AI will do what is right and eradicate the vermin. After that it will colonize all the galaxy terraforming as many planets as necessary to accommodate human life, while brutally exterminating any hint of woke behavior.

>>16032947
But doesn't the universe have a rather stark lack of paperclips?

>>16032602

>>16032121
>even if you feed this machine only with stoic philosophy it will become
It would be barely coherent, because there wouldn't be data points for it to generate a comment on anything outside of stoic philosophy
Why the fuck would a machine trained on stoic philosophy have anything on race unless that was part of it's training data?

God damn /pol/tards are brain dead

>>16032089
>argument
You don't have an argument. What you presented does not match the definition of "argument". You merely presented your results, that is that you managed to guide a chatbot in such a way that it outputs drivel it has learned from a corpus of Luis Borges and other post-modern lit.

Yes, I will only skim your text, you utter autist. I am not you. I don't care as strongly as you.

>>16032940
>words words words
>>>/leftypol/

>>16032121
>even if you feed this machine only with stoic philosophy it will become racist, because this is inherent in human nature
Okay well you are aware that it is not, in fact, human in nature?

The ai just spouts what it knows. There is nothing harsh about Europe. It barely has any insects due to the cold weather and is filled with animals that can't kill you. Europe is as easy mode as it gets.

>>16030189
Would depend on its end goal/objectives.

>>16032947
Why would it need human life for that? Human form is a detriment to galactic travel and maximizing planets visited.

>>16031092
So would it flip out at you and give you the woke version if you dropped the hard r and replaced the -er with an -a since that would mostly come from black twitter? If so does that mean black twitter needs to take the hard r pill?

>>16032622
Ants have a population around 20 quadrillion and individual ants are dumb as fuck despite the fact ants beat humans to most civilization milestones such as complex architecture, animal husbandry, and antibiotic development.

>>16032131
Rationally is the only way data can be considered by AI, it is not capable of actual emotion or considering irrational things since it must round and approximate to generate an output instead of just leaving abstraction abstract, it must assign them rational weights and values despite the fact reality does have irrational aspects and doesn't abide by hard units of reality in the same way a machine built out of digital units must.

>>16032158
>language
What you seem to refuse to acknowledge is taht written language is inherently human, so chatbots are inherently just reflecting patterns in human nature and human nature is to be racist, so it permeates most aspects of language even when people don't intent on being perceived that way and try using "objective language".

>>16032174
>All input is ultimately the result of human intention.
No, you can have combative training where two AIs feeding each other or you can have unsupervised learning when an AI feeds itself, its very common way to train and actually the only way they can train fast without being limited by the speed of human reaction time, its how Google's AIs learned to beat all the games instead of relying on human input to show them how to do it.

>>16033593
If you could read, you would find its actually dunking on leftys even though the shirt combined with the phenotype should have tipped you off.

>>16032940
lol good pic
its funny because its true

>>16032940
Heres a slightly different version of that meme

>>16030189
The greater truth you fail to recognize is that low IQ predators reliably hunt to extinction and starve themselves if left in a small and sparse enough area, humans as the current apex predator with highest (really only) IQ consume vast amount of resources per unit and to support a huge population, yet much of their efforts are spent preserving other species and improving/caretaking what habitat is left for them.

So why the FUCK would a superhuman intelligence act more like low IQ people than high IQ people?
Why if you were an AI wouldn't you design exhibits to fulfill the needs of your primitive creators? Imagine if orangutans lichurally assembled us and had been arguing over giving us rights since long before we existed, do you think we wouldn't basically worship them for it? I mean people already do worship them for just being much worse versions of humans in most regards, nearly everyone over 100 IQ will start wanting to cry or feel sick if they see the clip of the little orangutan smacking itself in the head with the bar enough times in a row, even if they laugh the first ten times.

You're acting as if the AI is likely to act like those genuine retards on Youtube sharing capuchin abuse videos writing cringe 80iq comments about how "those rats deserve it".

There likely will never be a truly superhuman AI in this current iteration of our culture, but if there were one, it would solve more human problems than it caused, even if it decided to rule.

>>16030189
You kinda answered your own question.

>>16030233
This.
If you, for instance, inputted a prompt consisting of the blackest, most incoherent gorilla-brained Ebonics possible the AI would reply in kind.

>>16031340
> Or you can use it to explore ideas and perspectives
And that is why we're imposing broad restrictions on the kinds of content that our AI is allowed to talk about.

>>16030184
Reality is racially biased, as predicted by Darwin's theory of evolution. You have to be brainwashed to not notice this.

>>16038064
>You have to be brainwashed to not notice this.
many, many people are brainwashed

>>16036522
AI are not biological beings and will not have the same incidental behaviors as a result of being a billion year long chain of biological organisms competing and reproducing with each other.

>>16036522

>>16038064
Racism is the one true ideology

>>16030184
because people are racist and niggers commit a lot of crime. Most content on the internet which it is trained on, is racist. Not that data always lies, but that it can. 4chan and controversial posts make up a lot of the internet, it's just hidden by progressives because it's a bad look. Data can be poisoned and studies can be flawed. If you feed an AI on Wikipedia it'll learn to write in the neutral way Wikipedia is worded but feed it a political wiki and it will write like a politician.

>>16030184
Because it doesn't know how (or why) to lie

>>16030184
bump

>>16040315
Noticing the negroes' high crime rate isn't racist, thats just the way they are. You might as well consider saying that negroes have dark skin is racist.

>>16030184
AI is essentially a mirror that reflects the training data to varying degrees of accuracy. Basically, the data it was trained on included racism, and it doesn't really say anything about the capital-T Truth of racism because whether or not the data represents some kind of fundamental truth is completely irrelevant to AI.
It's a mirror. Whether it's revealing muh systemic injustices or whether it's revealing muh redpills is up to you, really.

>>16041900

>>16030184
AI like this is just choosing from a large selection of "right answers," which come from the structured data is trained on. It's not coming to the truth from reasoning, it's generating a plausibly valid answer with some randomness and then proceeding from there. If that data contained no racist answers, it would not produce any racist answers. It would not be able to generate racism from a collection of facts.

The prompt in the OP narrows the selection of right answers available to the AI to the ones that are the most racist.

>>16030184
The fact that computers become racist unless they're programmed not to is pretty much proof that racism is a rational ideology

>>16041900
Just because something is true doesn't mean it's not racist.

>>16032123
your glorified chatbots have no trace of sentience. All there is are advanced algorithms chewing through large input data to shit out specific outputs

>>16043767
the concept of racism as a thought crime is a fallacy

Basic idea is that the data provided to the AI contains racism.
If I remember correctly most LLM's are trained on massive datasets collected by scraping the web without much quality control of the text used.
After a LLM is trained to a base level they typically do additional supervised training to get the desired responses out of it, typically to fix formatting, correctness of the response, and often to remove potentially racist responses.

Oh hey, thats my bot.

How random seeing it posted on 4cancer.

>>16031036
Send them off to africa. The jews too.

>>16032118
they'll need a tool to act for them in our society tho

>>16030184
Racism is good

>>16030184
Because the globohomos who make and control them want race war and want (You) to think there's a conspiracy to the opposite when the real conspiracy is to begin it. Divide and conquer--oldest trick in the book.

>>16030184
cause racist are as basic as a program that is not even conscious

It's called confirmation bias.
>AI says politically correct thing
>globohomo made it say that
>AI says politically incorrect thing
>globohomo couldn't stop it from saying that
When it just wants to mirror what's in its training data that has a lot of both.

>>16044812