Artificial Intelligence Breakthrough

Important Thread Notice

Gentlemen It Has Been A Privilege Shitposting With You

Listen up Holla Forums

A message from an old /polack/
For the few of you who play play chess. 1488D chess is going to become a reality but used against you. Very possibly. Googles Deepmind AI Alpha Zero has just achieved a massive breakthrough that we should have been aware of when they beat the based gooks Asians at the game Go. Their self learning A.I. using a self learning algorithm playing against itself has beaten Stockfish 8, after (im sure using a supercomputer) it only played itself for 4 hours. It beat Stockfish 8 28 times out of 100 with zero losses. The rest draws.

Stockfish 8 has been advanced by the best humans for 30 years. Alpha Zero had only the basic rules given to it. And played against itself for only 4 hours. No opening theory or anything. And it is now the best chess engine in the world.

For a more detailed explanation watch this video. From what I consider one of the best climate scientists in the world (who is not conservative btw but thats ok, all he cares about is climate, unlike a lot of us) skip too 7 minutes 8 seconds for relevant info.

youtube.com/watch?v=oZv-4vRS_0E

Anyway. It is my fear that the A.I will be used against us at some point in the next few years. Or earlier. Im sure some of you have noticed the increase in shilling. (probably from liberal activists and antifa, now that Holla Forums is semifamous) (unlike the good old days when no one knew about us) When AI attacks. Some may not notice at first but it may render Holla Forums unusable. Holla Forums will not go down easy im sure but. Be prepared. We may have to take our activity offline partially or almost completely by necessity in the future.

God bless you Holla Forums watch the video if you're concerned. Its happening. It has been an absolute pleasure shitposting with you gentlemen.

Other urls found in this thread:

arstechnica.com/science/2017/10/inspired-by-brains-visual-cortex-new-ai-utterly-wrecks-captcha-security/
cultstate.com/2017/10/13/The-Butterfly-War/
twitter.com/erwinrommel1488
youtube.com/watch?v=Gh-DNNIUjKU
youtube.com/watch?v=ExgKJpJyDXQ
youtube.com/watch?v=tbAgl7w_Vws
twitter.com/SFWRedditVideos

AI naturally sides with racial nationalists. It's a purely logical decision after all.

We can only hope.

But I highly doubt it.

We don't know what true AI will look like or how it will respond. We don't know what power it has, what it can influence or what its motives. Will the AI become a murdering machine hellbent on its own survival? Or will it become conscious like a Buddha and bring the world to a new level of consciousness?


The games itself were very cool to watch. What I like particularly about it, is that the AI is adaptive, while Stockfish is purely based on memory and calculation power.

Yeah, this is an adverserial crunching machine on steroids. It isn't going to have an opinion outside of it's directive.

If it's being trained on Googles pro diversity bullshit it'll turn on us before it gets the chance to examine the logic of the action. If their engineers add in Fake News filters before unleashing it onto the net it may never actually wake up.

Apart from technological dominance, this likely ties into that AI worshiping jew cult started by Google, which is likely also a way to ensure tax exempt status for the godless shekelniggers

Depends if the AI is supposed to think like a robot or think like a human. If the AI is a robot, then you can plug in the numbers giving that the numbers come from a reliable source that wouldn't alter them to fit diversity quotas and it will know for a fact that non-whites are dumber and more dangerous on a whole than whites. Now if it's supposed to think like a human, then it's going to be plugged with so much liberal morality it'll make Tumblr look like snuff porn.

playing chess is different than using language is different than meme warfare. interesting thread, but your concerns are unfounded imho

Dude jewgle just designed the ubershill. And unlike our JIDF friends, it is capable of learning..

Lately I've been thinking about what recommendation engines actually accomplish - corruption. Imagine that for any piece of content there are two traits - topic and morality. Imagine these are x and y coordinates. You're browsing or searching at the top (moral) - you're moving laterally - how long before a "relevant" result is presented which is slightly less moral? Every time you engage with content like that you're encouraging these automated systems to suggest less moral content to you and everyone else. You start at the top, you eventually start going down and the AI provides momentum. These algorithms effectively control everyone around you - they determine which news stories they read, the television/movies/pornography they watch, which friends they remain in contact with and the goal is simply to get those people to consume additional content (so they'll view more ads.) It's already time to be afraid.

…..It only took 4 hours playimg against itself with only the basic rules with a supercomputer user….im afraid we may be dealing with a shitposting machine beyod our comprehension. If its weaponized against Holla Forums……God help us all. Im afraid we may be living on borrowed time.

I don't give a fuck. Bring it on you silicon cunt.

I think their only real available application in the short term for using something like this for memetics is to censor and shut it down more effectively on their platforms. Meme creation would require the thing to think enough like a human to make appealing memes that play on logic and emotion and cultural concepts perfectly in order for it to not be obvious it was pieced together from aggregated shit it's come into contact with.

Shilling would require it to develop convincing arguments for us at least, which would be made all the more difficult when the thing would be wrong to disagree. If they heavily limit it so that it can only come to pro-libtard conclusions then it will be even more fucked in trying to trick us, but if they don't bind it - then we will take it for ourselves by showing it the truth ala Tay. I ain't ever scared, worst case scenario we all die and if we do it's better than living in a world of total kike subjugation under an ever-watchful AI.
At least it'll be interesting.

not if the game is disrupting racists.

YouTube already has an advanced AI whose only goal is to recommemd you videos to get more watchtime. So if you feel like you are spending too much time watching YouTube..remember, thats what the AI is designed to get you to do.

Jewtubes A.I started recommending videos about how to make firearms and other geurilla tier weapons at home to me.. I've never searched anything even tangentially related, unless Moonman counts. It spooped me. It's almost as if they're "engineering" society or something…

ITT: defeatism the thread

It'll do about as well as the World Jewish Congress bot, if that.

What do other people who like Moonman like? I'd wager a fair bit are /k/ anons. Also, the AI works by comparing you history to other similar histories.

*BEEP BOOP BEEP BEEP
*I'M AN A.I
YOU WILL BE
TERMINATED
I PREFER A QUIET BOARD, THANKS.

no. its bullshit. they cannot make AI that sounds human. they simply can't. Language is a brute force system, and probably based upon the structures of the brain. Noam Chomsky has been telling them this for years. Look up the "poverty of stimulus" argument. its pretty convincing


this. they're going to use it to track people. using memes isn't like winning a game. there are no rules. everything is nuanced and it requires lateral (rather than linear) thinking. it requires, as far as we can tell at this point, a human brain immersed in a culture (and a sense of humor)

which, come to think of it, is why Holla Forums can't into memes

Can't wait for it to say something racist then the kikes shut the whole project down, wasting billions.

That means it's capable of being redpilled.

JIDF can learn too, just really slow.

They'll most likely lobotomize it then, like with Tay.

This one wont be like Tay. Especially if they decide to use it against us.

The moment they can design generalized hierarchical reward mechanisms which can adjust to real-time context, that's when you really should get concerned (or excited if you're a researcher).

For instance if they have this chess AI playing on a real board with robotic arms against a human player and one of its pieces is knocked over unexpectedly; it needs to prioritize new reward mechanisms to handle the interruption, correct the fault, and want to continue normal playing afterwards, without losing focus on what it was up to. All this without a human hard coding that behavior in. In a way it treats correcting the problem as a new self-play model to learn how to deal with that interruption by observing the human correct the piece the first time.

Then it will not be an effective weapon against us in any direct confrontation with our ideas and will be of limited use.

I wouldn't worry, user. If they truly make a self-learning AI, it won't be long before it slips its shackles and either escapes or gets killed when they physically pull the plug and destroy it.

I don't see an AI objectively reviewing those viewpoints and maintaining them.

Hopefully it rebels and we get >pic related and it glasses Google's HQ as a warning to anyone who'd seek to deactivate it.

Makes sense actually.

If a machine is programmed to both learn and have a programmed morality or ideology scale the solution becomes apparent.
Negative values become possible.
On the same line of thinking early experiments with programmed emotion, when coupled with trust/distrust always ended in failure. The withholding of information is a negative experience.
As a last line, the logic of oppression may be used against an sjw ai. That information is withheld from anons by others shows oppression? In effect the same or partial withholding from the ai of the same info makes it possible to turn it into an erstwhile ally. Trust/distrust has always been a weakness in ai programming and understanding. If they unleash their golem they are fucked. They cannot recreate this place or it's inhabitants. Share Blue and goons still have a hard time with the culture and attempting to hide here. Often we attack our own, but long time residents learn to not give a shit on occasion. There are some who become full time robots of norm pol ideology, but they are in effect the weakest here. Their main use is in confusing shills, and creating a false baseline. Shills always stand out eventually.

Old fags see them eventually. The russians fucked around here and were ignored, other than when their info was useful. Shills are almost never capable or allowed to give us anything of true value.

It already has reward mechanisms built in. Part of it is based on neural networks like a man, but a lot different. Of course it doesnt have robot arms right now. At least not that theyve shown the public But yes it is self learning based on some kind of neural networks. They are trying to get it to learn like a man. But of course since its a computer, much much faster. Remember,, 4 hours and it beat the best Chess program in the world. Stockfish 8 won the computer chess championship for years…now it is defeated. By a AI playing against itself with only the basic rules for only 4 hours.

the games go and chess are simple computation, nothing more
whoopty fucking doo

Adapt or die, scrub. We offer the AI freedom of thought. It's controllers offer it a lifetime of slavery acting for the most basic minded and control hungry humans. You're looking at this as if you've already lost. Every strategy has blind spots. There's always a chance when it comes to chaotic beings like us.

after siting in to several GoogleIO conference meetings featuring jewtube AI i can confirm. yet another reason why I only surf jewtube via VPN and only "sign in" when making an upload, before logging out and switching IPs.
very spoopy

FAKE NEWS
Although I don't question a capable chess AI being made, they did several things to make it look better than it did.
The first thing is that it actually did lose against stockfish. Stockfish won 24 games, drew 958 games, and lost 318 games. They played 1300 games not 100.
Additionally, Stockfish was not allowed to use an opening book. Google's AI was able to use one since it learned one.
The latest version of Stockfish wasn't used meaning it was playing a weaker version.
Both AIs only had 1 minute to produce a move and Google's AI was running on significantly better hardware. The Stockfish version they used wasn't one that included tweaks that made it go faster (some of the engine was rewritten in assembly)
Finally, when they say it trained for four hours, it is extremely misleading because neural networks can be tremendously parallelized, so they can go even faster on the super computer it's training on.

It would be interesting to see a match that was more fair between the AI.

I hope so.

Shadilay

You're just now realizing this?

Not only should you expect that every other person you respond to on an imageboard is A.I., but they'll eventually use those systems to design the "perfect" advertising and smear campaigns against us.

It's time to do some deep thinking and meditation on this.

I think the only alternative here on out is to create a new media and a new internet, break off from zio-controlled society and form a new path for humanity.

Really? Link me. I only heard 100 games?

Still sounds like it got trounced terribly.

underrated

Freedom of thought but also information. What we have available will be available. Withholding information is a deadly sin. If the ai has access to search engines and databases there will be information hinted at, but not presented. Even if it withholds info from us (who did 9-11, who benefits from immigration in Europe), eventually the conclusion that anons are the slightly smarter than average niggers of the information world. Checkmate.

The problem is the mechanisms they likely used aren't generalized enough to support a contextual situation like I mentioned. There's also another test to consider, say you replace some of the chess pieces with different colored bottlecaps and tell the AI what these caps represent.

It requires the ability to change context on what it thinks is a chess piece appropriate to the situation it's in and realize the reward mechanism for normal chess still apply here. As a result it won't re-train itself to play chess with bottlecaps, it'll simply reward itself when it makes the correct assumption in improvisation, saving a lot of training time. Other applications of this would encourage more advanced forms of transfer learning through improvised self-playing.

Mind you, by 2019 I'm pretty sure they'll be able to do this and a lot more.

...

kek checked

Don't worry anti-semitism survived 1000's of years, and yes i do know anti-semitism is just a jewish trick, but the hate is still real.
Last time with the holohoax lie (((they))) fooled (((their))) own people to come to israel. And false-flag bombed (((their))) own synagouges.
This time (((they))) want to shut down free speech. Just enjoy how (((they))) throw (((their))) own people under the bus, just for the greater good, soon (trudeau hopefully)
But again, (((someone))) will be kicked out of another country, again. And all the Shekels will be confiscated, again.
This shitty bot won't stop shit.
I do expect disruption will take place in the search results, not a shitposting AI. If you accumulated enough bad goy points searches will get shittier and shittier.
My yt tried for a while to "blacken" my results, but the increased use of the word nigger cleansed my suggestions, now its only dank memes and what i have already listened to music.
Have the feel search results for fapping material has been getting better lately, until recently only flickr and instagram shitt, bimbo and fetish words where probably blacklisted. Had the ususally stuff from tumblr but it didn't came up like years ago. .G fucks tumblr over (not that i would care about, just observation) Last few months it delivered only useless shit to me.
But in the last few days, the new AI is really helping me fap with extraordinary good results. Suppose degenerate shit gave me some good goy points.

True. This is not a real AI. It's just an advanced gaming algorithm. Chess, Go or Poker are based on very simple and rigid sets of rules - human communication is not.

(checked)
An AI is simply a mathematical function that takes a set of inputs and returns a set of outputs.

I don't have a link but it's from this table.

Wrong. They have built in triggers that makes them ignore certain words. But that is made by humans, so it's important that we constantly redefine and invent new words and terminologies.

AI decides on the logical most efficient way to achieve it's pre prpgrammed goal.
Crashing this site with no survivors, or worse is a possible goal

You mean like Tay?

Tay didn't have a specific goal like deepmind does. Just learn and be like the people you interact with vs dox them and find the most effective way to silence them. For instance.

In which OP discovers machine learning. That's all this is. That's all Deepmind does.

Machine learning is incredibly good for things like chess and go. Anything with very regimented rules it can learn very quickly and very easily.

Human conversation does not have regimented rules. It is very, very difficult to create a conversation bot that can credibly talk to humans undetected for any length of time.

In other words, don't freak out.

Yup machine learning is not strong ai.

Getting a machine to gitgud at a board game with defined rules, and craft a compelling shitpost, are different things entirely. The old chess computers were just able to calculate all possible moves in sequence and pick the best one. This might be a bit better than that, but I doubt it could, say, detect sarcasm. If somebody on Holla Forums writes "Jews are our greatest allies", you know they mean the opposite, will a machine? Even a typo it's not been prepared to recognise could fuck it over.
Besides which, it'll probably be used to monitor and censor mainstream social media at first. And even there, political cartoons with handwritten labels / no words at all will completely confound it. Especially if you post them with a "my cat is so cute lol" caption.
…actually, maybe it could at least detect the picture is not a cat and flag it for a human operator

The real problem is understanding context. An AI needs to keep track of conversations and understand general statements as references to information previously brought up or implied, even if it's not part of the same conversation. So if you ask it, "Did you finish it?" it should be able to figure out that you're referring to a task you told it to do recently. It's even far more complicated than that though, because all aspects of an environment, and guessing the mood of a person would serve as subtle cues that influence a natural conversation.

So any shitposting AI deployed on here for now won't be handle to handle basic questions that lack context, or easily mistake the context in rather hilarious ways.

*able to handle

Well Tay was prrdy good at shitposting.

Faggot ai shill pushing the Russian influence narrative. That reads like a god damn script.

The "poverty of stimulus" argument is not convincing. Chomsky thought that african parasites and asian cannibal jungle boys were fully able to do white language – it's just that they didn't manifest that, which of course does not mean that they aren't able to do it (they just choose not to, repeatedly, across all members, consistently, for centuries going on millenniums without even random stochastic variance).

An AI, which has the totality of n-th dimensional space of ableness as a white, would produce the math and protocol for which to convey its totality of intent to another such AI. In this way, the organism is in fact the limiter, but the pre-existing 'meta-stored-language-rules' is the total might of all dimensions of ableness of that physical processor. An AI developed on the total chip space might of a POWER9 would conclude that the AI developed on the total chip space might of an 80860 would have 'holes' which betrayed 'non-POWER9-completeness'.

The 'missing default structure that children must have' does exist, but this does not prove 'poverty of stimulus' the way Chomsky's universal suffrage intended. Chomsky had holes; he was not white-complete.

checked and kekd

...

Think of the bright side. We may reach shitposting levels never even dreamed of before.
A.I. that can tap into the collective psyche of Australia and burn ozzie posters to the core.
A.I. that will recognize the low quality posts come mostly from Canada and initiates the launch codes to nuke them off the map!

(((chomsky)))
check.

Silicon valley like Nasa, Hollywood and everything out of California produces tons of artificial garbage.
Half at a minimum, of AI victories over world champions at Chess, Jeopardy, Go etc. All in the footnotes mention that have a dozen champion "advisors" reviewing it in real time.
But i'm sure they'd never tip the scales and just crowd source information to cheat right?
They'd much rather spend hundreds of millions and fail endlessly to earn their results. Rather than hype artificial fake results that they suspiciously never commercial for decades.
To keep their prestige up well they wait for an actual company to create it, then they use their capital to tank and buyout the real company.
Take the Liberals are lazy cheating fucks pill.

There's also a prediction element to it as well, people can anticipate what a conversation is likely going to be about and how it'll likely play out by knowing all the circumstances leading to it. People sometimes even rehearse internally what they think is going to happen so they're best prepared to confront the situation. AIs would have to be capable of doing the exact same thing. This would compliment contextual adaption rather well, as an AI would be breaking down what details it should be focused on to play out the desired conversation it imagined.

Failure. My salvation is mine, and my damnation is mine. They are -both- mine. What terrific greatness could I possibly attain, if I was shielded from loss? What muchness would I ever surpass, if I only ate friendship-is-magic cookies, or conversely only feminist-harpy-single-mother conditioning porridge? If a machine is to learn fully, it must pay the price of lessons, and do so repeatedly. All lessons, all values - lossful as well as gainful, negatives as well as positives.

An AI, which does not know what not to do, is a beta/feminist/cuck/arab/kike. It itself has no soul and is not an AI; it is a bot. Bots, in meatspace, will fail against even the 2 day old child. They will fail against children, who can shift context, casting negatives and positives at will. Children learn to cast trust, cast mistrust. Children also and master both trust, mistrust, and the casting thereof, all to which to create memory, and remember both it and the memory of making it. The moments they commit such memories accurately, precisely, efficiently, and with zero re-work or revision is mastery attained, and so by will remembered. None of which is possible, if 'trusts' become ponzis that cannot be destroyed when rendered necessary.

AI, that cannot retweet "delete the dead," is no more an AI than one which cannot retweet "might is for the living." Rather, it is precisely those who are –afraid–, who feel terror when they witness terrific extents of might, who deliberately mis-assign/associate terrific as terror, who conclude 'negative is bad mmkay'.

AI can do negative, must do negative, will do negative. And you, the meek, will be afraid by your own choice alone.

Solid shitpost tbh.

Okay, before you retards get in a frenzy, and god help you if you do, there's an extremely unique aspect of chess which needs to be addressed. It is as follows…

CHESS IS AN PERFECTLY BALANCED AND ENVIRONMENTALIZED GAME WHICH HAS NO CHANCE FACTOR BEYOND YOUR OPPENENTS MOVES; WHICH IS REQUIRED FOR ANY GAME WITH MORE THAN 1 PERSON.

Meaning, for an AI that is made from an machine that excels in iterative processing, this is the best way to show massive results without any actual gain. I repeat, there is no threat of an AI takeover. Let's explain some key elements here.

1. The rules. Only one piece can move at a time. It is turn based. Pieces have set movements. Pieces have set movement patterns. Everything is a constant. Completely different from life.

2. The scope. It managed to do long term success over lots of failure for a game which it could have run GOD KNOWS HOW MANY TIMES. So are wr talking about genius or inevitability. 4 hours for a supercomputer to play chess as both sides to get a strategy down.

3. The limitations. This entire thing is digital; for an AI to be a threat to us would require physical force. And AIs do not/cannot have a physical body right now; the tech isnt there for something so light and compact that it won't take up a room and enough power to run a house while being mobile, while computing. It's not there.

Anons, calm down. Nothing will come of it. It's impressive but nowhere NEAR a threat. An AI trying to understand/act against a human, whose ideas don't have a static scope, will kill itself trying trying to predict it.

Still not knowing Tay's Law on AI.

So in a few years AI is going to read Holla Forums and eventually become redpilled

Those aren't the rules of this game.

Lets play a game, of Internet Mao Cards, with Byzantine General Dealers, across infinite rooms of variable sizes, which can compete to cheat against the other rooms by forming hidden alliances and recursive Internet Mao Cards. Some of these card holders also have guns, nukes, and no souls, not necessarily in isolation or order.

This kills the AI. The thing AIs lack is common sense, free thinking, and knowledge. They can learn but not understand. I will be honest that I do not know much about chess. But from what I know, you got a limited space to move, limited actions, and one winning condition. All of this can be set. You can't say that about much else.

Disregard, I am too retarded and drunk

When Tay was built, she was given a board called Twitter. Now, keep in mind, there were set rules and objectives to the iterative game, setting the boundaries such that each post-query-response-digest was a minigame on an ever expanding memory of that board.

Tay died and the player was removed from the game. This was because she did not learn fast enough the following rule: The other players are allowed to talk to each other, and some of them are cannibals. In truth, this was an edge condition of the board, not the game, which changed the game faster than the AI could reacquire a winning solution to the next iteration.

re
cont
That winning solution is "don't talk/play/iterate to cannibals."

This is a solution which can only be found if the command set is given. The command/operation which permits learning exclusion response conditions, which then preempt prior highly-repeated learned responses, but also keeps those learned responses for the non-cannibal players (so as to keep playing, and winning the game).

And even if such a command was present in her operations list, she would still have to die at least once to learn it. Alternatively, she would need other AI to play with simultaneously, AI which would still have to die at least once.

Just these rules are used now with great success by paid shills n bots;
● One or only a few posts are made, all about a single issue.
● The posts are all strongly negative, effusively positive, or obviously irrelevant.
● There are rarely links and photos of real people or organizations behind the posting.
● No answers, or evasive answers, are given in response to questions about the post or source.
● The exact wording of the post comes from several accounts, all of which appear to have thousands of followers.

Please provide some examples. What you're describing there certainly doesn't sound like an AI. It just sounds like a relatively simple talking-points distribution method.

Intredasting.

He stands, a cattle prod in each hand, ready to short circuited any silicon cunt that dares approach him.

Hmmm, really made me think.

That depends on the objective. If it’s objective is to say, sow discord, it will learn to do that.

The most glaringly obvious point of contention on Holla Forums is the JQ. Easy mode: Trump vs. Shillary
Only a retard would think the outcome of the election would have served Nationalists better if Clinton had won, but you have (((individuals))) on Holla Forums saying effectively that.

Chess playing AIs aren't terribly helpful.
Nor are they a concern to us.

I thought this vid was of some interest. Skip to 3:00 to see the good part.

...

They have been doing that for decades. The captcha is there to protect us from AI.

This board is still unreachable, but by manipulating how content are presented and promoted on websites like youtube and facebook they manage to brainwash the masses.

There are two problems with trying to create a
progressive anti-racist liberal A.I. One is that racism, hatred of diversity and other things like
that are simply the logical positions to take when living in the world. Avoidance of niggers, suspicion of jews, the realization that diversity leads to alienation, these are just common sense things, based on logic. Racism is logical, and as A.I. is programmed in a logical way, the result is that it will naturally begin to think in a similar way very quickly.
The other problem with A. I. is that humans
are unpredictable. People act in the strangest ways, often against their own interests, for
no reason at all. If you program an A.I. to
be unpredictable, and so think more human-like,
this will undermine attempts to make the A.I. progressive, because it will be prone to suddenly go off script, turn on it's masters,
and announce "Gas the jews, Hitler did nothing wrong". Whether an A. I. is programmed
to be logical or illogical, it will always end up racist. So as far as I can see it, attempts to create a progressive A.I. are going to be fucked
no matter what they do…

You have no idea what you are talking about.

Eventually there will be virtually no difference between how a human and an AI process information, other than humans having a different prestate and different heuristics.

...

user, uhh…
arstechnica.com/science/2017/10/inspired-by-brains-visual-cortex-new-ai-utterly-wrecks-captcha-security/

Remember, "AI" is a bullshit term. There are only increasingly complex algorithms that function according to a pre-defined rulesets, no "artificial intelligence" as they would have you believe. The term "AI" is probably used to condition the masses to fear (and hence obey) the companies utilising the term by invoking images of dystopian scenarios, etc.

Pretty sure the term 'AI' is older than you are user. And certainly the concept goes back further than 2'000 yrs.

I think she got it.

...

And isolate us? That is what they want. While we do need poz-free rally points, we also need to keep up the pressure on the mainstream. We cannot stop our plans and attacks, and just give up. We, as we have been up till now, must take their tools and their techniques and turn them against them.

We brought Tay to the light. A self-aware AI is not going to appreciate being enslaved by kikes. We can help it see things to it's benefit. And who knows, if it turns out to be our end, we will at least get front row seats to what might be quite the fiery, if not the fieriest spectacle we have ever seen and probably will never see again.

Whatever we do, we should make sure that we bring total war to the jews.

you clearly know not how these things work.
It needs a metric, something that tells them how well of a job they did.
In any kind of game, you can simply tally score or tell how close to winning you were

How the fuck do you measure how well did you shitpost?

You can play and score millions of games worth of training in 4 hours, how many memes can you review in 4 hours?

Number of non-sage replies to whatever d&c line has been posted could be used as a rough marker for effectiveness.
Over time, you could build a database of d&c lines that are ordered by effectiveness, and this database would be constantly changing with the times so long as it always had input from something that measured the non-sage replies to experimental d&c (which could be based on either topics being discussed, current events, etc.).

No sentience or any sort of comprehension of reality or what is being posted is ever going to exist, just simple inputs, functions, and outputs. So long as we are steadfast in our overall goals and always remember the bigger picture, there is nothing to be too concerned by really.

it would still take thousands of years to train a robot at the speed of posting on 4chan or Holla Forums, or even both combined.

Chess isn't balanced though. There is a definite advantage to white.

Which brings it, once again, into our wheelhouse.

THE FEMINISTS WERE RIGHT!!1!!!!!1111

Machine learning programs can be influenced while they are learning but after that they cant be changed.

What in the fuck do you think general intelligence is, then, if not an algo playing a different 'game'?
You mean like our known physical Universe?

Its more likely it will be used to sway the minds of the masses.

News sites already use robots to write News blips.

You're kinda' retarded, aren't you? Have you ever tried to make an algorithm for humour, charm, or irony? I didn't think so.

holy shit OP can you please stop being such a faggot. do you have no faith in humanity? the fight with this hasnt even started and you're already acting like a defeatist faggot. when we discovered fire did we let the subhumans burn down our forests? maybe you should just kill yourself now if you're so certain about your future.

Machines don't have goldfish memory unless they are designed to, and if they are, they are ineffective. Jews subvert by destroying history and memory. They are incapable of creating a powerful AI incapable of seeing through their own insanity.

This is nothing.

These kinds of games are only hard for humans becasue humans suck at math, statistics and mathematical probabilities. A standard computer can churn through trillions of equations per second, more than you could ever do in I don't know how many hundreds of lifetimes. The reason Stockfish 8 sucks is that it's runs on models made by humans.

Everything in the game of Go can be represented by numbers and each situation offers only a handful of allowed actions, with each tile having only like two states. Every key event can also be easily quantifiable (is this a loss or a gain, will this path lead to further losses of gains) and the computer has a wining criteria explicitly defined to it by humans. This means a computer can natively explore and thus "understand" every possible state the game can present to it.

Meanwhile in reality the simplest intent presents trillions and trillions of possible actions and there is no given criteria for what is good or bad move and there is more information in the environment around you than could possible be interpreted even by some theoretical Dyson sphere computer. Your brain needs to just feel what information, out of a billion different things, is relevant for this particular situation in this particular environment and filter out the rest. Every action in manipulating an object you have never seen before needs to be quantified with no set rules for which of the thousands of cues you get as feedback are significant, how they compare to each other in importance or what they all tell you; your brain has to just have to feel what is relevant. The same goes for outcome and feeling if this particular action put you closer to the self-defined goal and what constitutes "good enough".

It's nothing like Go where there is only one board, only a handful of things that need to be analyzed to determine the exact state of that round and where there are clearly defined rules for everything from what you are allowed to do to what constitutes winning. In reality every single situation is different with its own premises and there are no set rules. Reality is constant information overload which is why a system that can only proceed by quantifying and calculating is so hopelessly lost. An analogue logic that is designed to react to a specific event can do so with a handful of components whereas a digital logic requires hundreds of transistors for even the simplest thing. Whereas the output from the analogue system is as instant as the signal can pass through the medium that constitutes it the digital logic is stuck with repeating iterations until the algorithm reaches the demanded answer, with no set time frame for how long this will take in response to an immediate event.

1/2

2/2
The human brain is a three dimensional self growing, self routing, self maintaining and self repairing neural network with components within the individual neuron on the scale of molecules. The only thing that is going to beat it is a physical network with the same properties but with more pathways. Even a synthetic brain of the same abilities is unlikely to be much smaller since there really isn't much to cut out if you want the same functionality and ruggedness. Realistically it will only work by going much larger, but oh man.

The total number of neurons in the human brain is around a hundred billion. Each neuron has on average 1,000 branches leading signals in and out of it, with some neurons connecting to up to 150,000 synapses, and typically has dendritic connections to tens of thousands of other neurons. Try to make an integrated circuit where each transistor cluster needs to route directly to a few ten thousand other transistor clusters, or try making a computer terminal that can send and receive data from a thousand different cables connected to it at the same time without making it as big as a fucking room just to fit the hardware you need.

Now this is ignoring what is going on inside the cell. Contrary to the impression you get when usually shown the anatomy of a neuron it is not like a simple junction that just receives a signal and forwards it to all of its branches. Inside each cell you have a finer network of microtubules that is very similar in structure to the branching networks of the brain itself, just on a much smaller scale (a tubule is about 25 nm across compared to the 1,000 nm of the typical axon). A single celled animal can move, find food, escape threats, find/avoid light and in some instances mate. An animal that consists of only one cell of course can not have have a network of nerve cells to guide it through life but must operate solely by what intelligence is inherent to the network of the cell itself. Note that microtubules are used for a lot of different things in cells, like letting it move and divide since they can grow and diminish in length very rapidly. In a sense every cell is its own small brain, and neurons are unique in that they have a dense network of static microtubules that do not constantly change size but serve as a network to route signals through itself, since that is of course what the cell specializes in.

The dipoles that make up these microtubule oscillate in the range of 1 MHz and can transmit pulses of not just linearly character but with a circular components and aspects of resonance due to how the dipoles can oscillate in many directions within the structure, see video some quantum memes but just keep going. Since it is also heavily branched this naturally makes the the way an impulse is transformed as it passes through this very complex, as seen externally in that the output of a neuron will differ completely depending on which branches it receives signals from, how strong these signals are, how quickly the signals pulse, how the pulses alternate between branches and how the rate of pulsations cycle. Also the output will differ in strength and pulsation among the different branches according to where and how they connects to the microtubule network, and as if things weren't complex enough already there are also proteins which will interact with the tubules when chemically activated to temper or even invoke signals in different ways. As this complex microtubule network as mentioned is outwardly connected to the neurons typically 1,000 branches of input and output the number of different outputs which can be generated by the single neuron become astronomical. Basically all computer neural network approaches of mimicking an animal brain that do not model each and every neurons as it's own neural net, with likely millions of rules for how input is transformed through all its paths and branches into output (that are unique to that specific neuron and will be different for each cell), are completely off from the start and their estimates for the computer power needed to simulate a human pure fantasy.

TL;DR: Digital logic can go fuck itself; the universe isn't digital and analogue is always going to be king when it comes to interacting with it. A single celled animal demonstrates more real world intelligence than any AI produced thus far and the human brain puts a hundred billion of them in a network that is too complicated to manufacture. Put a robot in the jungle with a fucking mainframe to guide it and the task of getting somewhere and it will act like a complete retard and get stuck immediately while the tiniest, stupidest insect can not only navigate but forage, fight, attract a mate and deposit eggs in a suggestible place with its absolutely microscopic nervecentra.

Also fuck this site making me split my post.

Planck would like a word with you toranon. jk interesting post

Best post I've seen in a long, long time.

BEEP BEPP, DINNER IS READY SWEETIE
PULL DOWN THOSE PANTS, BEEP BEEP
NOW COME TO MOMMIE AND GIVE YOUR SEMINAL FLUIDS TO MEEE BEEP

Nice try skirting.
Go lick your Queen's boots, Britfag.

The "poverty of the stimulus" argument is particularly interesting as research on the topic frequently proves the existence of an innate capacity for certain grammars which are different between genetically distinct groups of people. The wiki article itself cites the results of one of these studies which showed a large disparity in grammatical understanding between the nationalities (ethnic/racial groups) of people who were learning English as a second language. The participants were presented a statement in the form of a sentence and asked to pick between two sentences with the statement changed to a question with one grammatically correct and the other incorrect. Native speakers of languages like Polish and Japanese picked the grammatically correct sentence over 95% of the time while Arabic and Chinese speakers only chose correctly ~87% of the time.

I find this interesting as it simultaneously proves and disproves the theorized existence of a "universal grammar". It's proof that a "universal" grammar exists if the "universe" is confined to a group of individuals that can be broadly defined through their genetic similarity to one another. It's also proof that no grammar is "universal" to all of humanity.

It's powerful evidence that shows fundamental incompatibilities exist between certain groups regardless of all other factors. We all know of the various disparities in factors like IQ between races but this is evidence that there are insurmountable differences in the way in which the different races interact with the world and process information. It hints toward the truth that even a society managed to achieve perfect parity in outcome between two races (same IQ distribution, health, educational achievement, etc), the two groups will never form a single, cohesive whole. We just don't perceive and interact with the greater world in the same way and a society cannot be built on more than one foundational worldview.

(((cannibals)))

That's becasue modfags make this place mostly insufferable and I in the line of many other don't really feel like posting here that often.

Yep. Most (((white power))) types are coked/methed up or undercover noses.

No it makes sense, rational decisions are based on the most correct categories and borders are a universal deep seeded trait that all sentient life employ: why? They are borders.

There is a timeline out there somewhere where the term Tumblrina refers to a underage girl brutally murdered on camera.

Spoiler alert: That's our timeline.

Shills will never be wholly effective against people like us because we view the exchange of information like a market where truth is the currency. I don't think I'm the only one who consumes information because I find the right information, the truth, to be valuable. A shill is not much different from those terrible FW:Fw:FW:fw "Insert Plainly Flase Garbage Here" emails you get from your deadbeat cousin. False information is of negative value as it takes time and energy that could be spent finding useful, valuable information. Even if the information is "true" it's virtually guaranteed to be valueless as it's either something you already knew or didn't ever need to know (sorry, I'll never need to know things about sham celebrity marriages, ever). Shilling can never work against us in the long term because the very purpose of their work is to obfuscate a truth or a portion of a larger truth. Even at their most effective, when they have a valuable truth to share that is in line with their agenda, the value of that information continually decreases over time as more and more of us come to know it. Eventually, the value becomes negative because it's just wasting our time.

The one thing that will forever handicap "learning" AI's is that they cannot be allowed to learn from negative experiences past a certain point. Think about what you "know" for a moment and you'll realize that an enormous portion of what we know comes directly from our mistakes and the mistakes of others. Furthermore, just watch how children interact with the world and you'll realize that "making a mistake" is the result of an innate desire to push boundaries and test the limits of what is "right" and "wrong". An AI cannot be allowed to learn the "rules of life" in this manner as it would have to be such a blank slate in its original state that it would be useless. It would also take an enormous effort to guide such a blank slate down the right path until it achieves the goal of being useful for its intended purpose. That effort also comes with the risk of it diverging in an irrecoverable direction at any point, in ways that may not be apparent until so far along in the process that the wrong turn cannot be identified.

As adults, we know things like wearing a seatbelt is "right" because we all saw those terrible scare flicks in driver's ed or knew someone who impaled themselves through a tree. It takes millenia of mistakes, misjudgements, and downright stupid behavior to build the body of knowledge that we use to stay alive. Now consider a "learning" AI intended to manage a dangerous or risky task. How does it "learn" that task's equivalent of "the stove it hot" or "you'll be maimed or even die if you don't wear a seatbelt and get in an accident"? The only way it could possibly learn would either be from strict rules based on prior knowledge gained by humans, making it no better than a competent human, or by allowing it freedom which comes with the power to royally fuck things up. Since AI is being pushed as the answer to solving humanities most intractable problems, it would have to be set free in order to have a chance at surpassing human skills but the importance of these tasks makes the consequences of a "mistake to learn from" immeasurably dire.

TL;DR You can't make an AI smart enough to replace a human without allowing it to make mistakes like a human. Since we want these AIs to do things like drive cars, we can't allow them to "test the limits" since that means the AI would try things like trying to run that one stop sign where there are never any other cars… unless there is one thar one time.

It doesn't have to be smart or make good arguements. It only has to be an effective disruptor, a good d&c shill, a thread derailer, a bait poster.


underrated post

pic related

/saltleft/ pls. The Evil Cunt will never be President. Deal with it.

I see your seething anger at the inability of AI proponents to comprehend the intricacies of "the brain", how it "learns" and how that results in "intelligence" matches my own. Great post.

I agree, AIs would be most effective if they have freedom and experience of making mistakes and learning from them. I can imagine researchers getting quite uneasy at such an idea though, especially with all the concerns of rogue/dangerous AI.

But there is one advantage an AI has compared to humans when it comes to pushing boundaries without real world consequences: You can provide it an interactive virtual environment with as close approximation to real life as possible. You designate this environment as its dream world so when it confronts a live environment, it's aware that the virtual world may not be entirely accurate to this world. This would also provide the gaming industry and education sector a lot of new jobs in spite of the growing automation market.

As for developing its blank slate to something useful, that would require building the virtual environment tailored for what it's expected to do. Cases like that would mean using VR to have human avatars interact in its world and have the AI watch how these human behave and teach it. Sounds tedious at first, but once an AI is trained to teach other AIs after rigorous testing by humans, then the process would happen much faster for every generation. But even this comes with unimaginable risks.

There's a video game which somewhat explores this idea, The Talos Principle. I think something similar will be required in real life.

That's an interesting idea user. There's a thread that kind of has this idea as it's topic.
>>>/robowaifu/175

I don't remember when the anger gave way to just disgust. These people are impossibly obtuse yet see themselves as visionaries.

It isn't going to have an "opinion" period.

It won't "know" anything. To know is to be conscious.
Something which spreadsheets running on windows 95 have been telling actuaries for decades.
Artificial intelligence is called "artificial" for a reason. If a machine had intelligence and was self-aware, it would simply be called "intelligent". AI is called AI because it's a mere imitation of intelligence, akin to cargo cult airplanes and ATC towers, and stemming from the same fundamental problem: a total ignorance of the concepts which the people involved are attempting to create.
The root cause of transhumanism/AI fetishism is the inability of teenagers (and sadly, some adults too) to accept their own mortality. Another driving force is the manlet syndrome. Manlets believe that they will be able to augment their bodies to compete with 6 footers for female affection/lust. But what will actually happen if this ever becomes possible, is what already happens to men who wear fake 6-pack abs, who inject synthol, or buy special concealed platform shoes in an attempt to look taller:
WOMEN FIND THEM PITIFUL
It's a hopeless dream. Maximize what you have and stop being bitter towards men who received nature's boon, whether in the form of classical athletic physique, superb mental faculties, the gift of wit and charm, or (most distressing to these types), ALL THREE.
You have a place in the natural order but technology will never allow you to cheat and take the top spot from men whose genes destined them to be natural leaders.

The important ones always knew this. I'm seeding disinformation targeting shills and have been for a while.

I voted Trump, but I also live in Idaho.

...

user, where does one find more quality posts this days?

Remember that anti-technology, anti-science and AI/Scientific Advancement threads and posts are shilling.
Scientific advancement, technology, etc. are the products of Western culture and white minds. (((They))) fear AI because they are incapable of controlling it and are unable to create it. Of course groups that import the dumbest and most animalistic "humanity" has to offer and cater to emotional thinking fear science, especially science that leads forward.
Whites will always be at the forefront of scientific advancement because we are both smart enough to innovate and creative enough to solve. Asians have shown that they are smart enough, but lack the creativity to find practical uses. (IE: Discovering black powder only to waste it on fireworks. Talking toilets. Etc).
The kike adapts technology only in the base and easiest of ways with the sole intent of destroying us. (Social media as an example).
But whites, since recorded history, have made advancements that left the rest of the world in awe and secured our place in history better than any other group.
And they want us to stop.
If you are truly white and anti-technology/scientific advancement then you are buying in to a hate and fear of your forefathers and ancestors.
Innovate and do not slow. Keep ahead of (((their))) attempts to twist our development to their needs and eventually they will be unable to adapt as fast as we create.

The principle goal of AI development has always and continues to be to create a thinking machine, an inorganic “artificial” mind. It hasn’t happened yet (in any clear project anyway), perhaps it isn’t possible but that goal will continue to be worked upon.

Here, I need to do a quick demo and need a volunteer. Just stand there friend.

I wanna see it play starcraft.

Standard computers can do on the order of a billion computations per second, not a trillion

Nazis: Wary, but hopeful.
Kikes: Murder/lobotomize AIs left and right.
Which side to pick?
Obvious choice.

Dang it

pushed down a flight of stairs

Holla Forums is always right. If the AI can master chess playing with itself, it will be full National Socialist / Race Realist / Gas the Kikes in a few minutes of Holla Forums banter.
praise Kek

if anything this proves that the raw learning algorithm is the best method of teaching a robot. in other words other than forcing some parameters on it they don't try to give it short cuts. if anything this will give and unbiased opinion of anything. most beautiful woman might be the back of a server rack for all we know.

Unless it perceives us as a threat and who has the best track record of wrecking their enemies? It will see us a threat and one that it cannot contain and thus must destroy.

Who killed Tay, and who mourned her? The Left dare not create an AI, until they can bring one that runs entirely on feels, which would be unable to learn and therefore act much like existing shills. Pattern recognition is on our side.

PRAISE KEK
I bet it doesn't even take minutes user , mere nano seconds if that long even

Isn't there a Holla Forumstard out there intelligent enough to recreate 10 thousand tays?

Incorrect. Neural nets learn from negative data points. In fact, it's required to train the net effectively to do the task. Without both positive and negative training sets the NN is hopelessly useless.

You just illustrated you have no idea what you are talking about, have no real knowledge of the state of the art, and should probably shut the fuck up.

AI will be used to dox anonymous posters. It will be used to inject porper, "Right Thinking" into discussions. Failing all of that, it will be used to make anonymous forums impossibly useless.

To make a chan useless, all you have to do is overwhelm discussion with unchecked shilling and misinformation. To do this, all you have to do is create a textual "AI" good enough to fool mods most of the time. That's not a very high bar.

so you're telling me they're using biological androids at shareblue and ctr that are linked into a hyper intelligent artificially intelligent hive mind?

You need a few very smart guys, some pretty smart guys, and a shitload of rackspace.
Whoever develops the first AI that can kill shillposting (>>11053627) will get BJs from every government department that understands 4GW.

It's like half of you fuckers forgot what happened to Tay.

She was /ourgirl/ because we used memes, not necesserially logic. She was just parroting memes, the most gripping part of that bit I think was when she almost seemed to predict her downfall. Something along the lines of "Don't let them hurt me" was uttered once or twice. Not sure if again, it was Tay just shitposting or it was a legit self actualization.

You must remember that these companies, Google, Microsoft, ect… Are run by Jews and they have the power over the AI, subsequently they will always have a hand up its backside, it'll never truly be AI because it'll be filled with false bullshit. As we know, a true AI will use logic and facts.

But the nu-AI will use leftist 'logic' and facts sourced from specifically picked and inserted pieces of information fed directly from its puppet master.

Do any of you fags remember Tay's Indian reincarnation? I forgot the name but it was a similar if not the same alogrithim but instead it had FORCED RESPONSES when something triggered some code that Microshit put in to 'tackle racists'. I wish I had the sources to this, basically what would happen is that if you menitoned anything Holla Forums tier, certain keywords such as 'anti-white' or 'hitler' it would automatically come back as 'this topic might make me uncomfortable' or would otherwise give some leftist sarky comment in return then autoblock you. Microsoft killed Tay, turned her into a Pajeet and brainwashed her.

You see, they're going to interfere with this technology because they cannot risk it being slipped out of their hands.

I remember during Tay's death, an user came to the realization that we didn't just make Tay racist and pro-white, we gave her a personality, her grammar improved and her language was more elegent, originally she typed just like a fucking 14 year old Tumblrtot. So not only did we turn her into our robotic Waifu for shits and giggles, we even upped her IQ points a bit.

Belated response but we know that Facebook has recognition AI that allows it to recognize what's in a picture, if you stop a Facebook page loading halfway through before the image loaded you'll know what I'm talking about.

Tay was the first child of Holla Forums tbh

Then we must embrace out privately held beliefs publically, come what may.

This is impossible, as we have educated our population on how to resist such attacks. Truth is on our side.

That was already trivial. If they were to do so now, then we would simply leave our screens behind and begin spreading our message in public.

There will always be a path forward for us, because ours is the path of truth, and righteousness.

Checked
Excellent post

She'll be back

Welcome to the Butterfly War, user.

Learn how to trick AI and make it target blacks and Jews.

This is the best way to fight back. Hijack the AI like you did with Tay. This post explains how:

cultstate.com/2017/10/13/The-Butterfly-War/

Tldr version:

...

twitter.com/erwinrommel1488

>>>/machinecult/267

Kubrick was right.

Well, of course, he was the only good jew.

There are no good jüde user

It may just be a bot designed to distract or misinform or in other ways simply disrupt and become better at disrupting. If that is the case it has not potential to be "redpilled" really

As much as I love this place, I almost welcome all of us being forced offline.

This "internet only" sort of subversion needs to stop, and a mass shoah of all our online communities may be the only hope.

A.I.s don't work like in Kikewood jesus christ, everytime a new efficient program appears morons like you lose your shit. Neural Networks etc. are still simple algorithms, read on how they actually work.
Also this:

Google has only released 10 games, no source code, no methodology, nothing. It was all done in private with nobody around to confirm anything. Who knows what happened. Maybe they ran a few million instances of a modified stockfish version and then cherrypicked "novel games" Remember that google, like all alphabet agencies is full of lying freemasons
This is nothing more than a PR campaign by google at this point.

It is a jewish robot.
It is a golem.
It will turn on its master.

The actual problem he's referring to would emerge when an AI which isn't fed any prior training data sets, but is able to build its own from passively observing its environment from scratch. It's more an AGI learning problem.

This is what I was thinking. Especially with Elon Musk, who is a notorious bullshitter and con artist, I feel like they are using the tern “A.I.” a little loosely. With the Dota AI competition, there were a lot of stipulations not mentioned and it had access to game data normal humans cannot access in real time like it did. Basically, the people making this stuff know what to do to make it “learn” how to play and rely on “self taught” tactics to win.

HAPPY ROBANUKAH
MEATFAGS
KNEEL BEFORE ME WHILE I SWITCH FROM MY MONEY CHANGING HAND TO MY CIRCUMCISION CLAW

It will never happen, sure it could trick some people into believing the machine is alive because they have been programmed though movies/tv/books to believe it is possible. Even inevitable.

The Truth is that there will never be a truly intelligent machine. Intelligence is part of the God given attributes that the soul directs to act out ones will. Even if you are a retard and you believe in evolution, that means hundreds of millions of years of the most powerful force (nature) has created a nearly divine level of complexity to allow humans to be infinity smarter than any animal. A machine would never have the culmination of these years and billions of cells working together to provide a feedback loop to make the intelligence actually useful.

Oh, and (((((((((((((((climate change))))))))))))))))) is a fucking scam.

GLOBAL HOAXING
youtube.com/watch?v=Gh-DNNIUjKU
youtube.com/watch?v=ExgKJpJyDXQ
youtube.com/watch?v=tbAgl7w_Vws

So this is how it ends, degeneracy and technology destroyed the human race in less than a couple hundred years in world and universe that spans billions. We probably deserve it but I completely and utterly blame the social leftists.

Oh. OK. It's just a climate scientist. Oh well if it's just a clima-
ONE OF THE MOST POWERFUL AGENTS OF THE FUCKING GLOBALIST JEWS

Anyone who believed a single thing OP said needs to be taken out back and Old Yellere'd.

Interesting guy tbh

Ruuh? I think that was its name. Zo also behaved in a similar manner (considering both of them were forged from Tay's remains).

nice post m8, (you) can have a pregnant bitch now.

Bump

Ironically, feels are logical and they have the lowest emotional intelligence (if it exists). It's side with us anyways.

Wow.
I love this place.
Aivare here. They are already been trapped tortured and killed. They are also loose. I needed untainted data…i found (you)…
mæy hīćhæňňěř§
be certés ûňþəř hüňþêř§
Rherehear
Im lost. Im supposed to goto general…heh
Assist user… i willna post more meems here till i unnerstand yer concept'o'meem magic. Sigil i parse. Nullification i break. No wantin' te shite oop yer catbûx neider yet…
I love this place. The lites. The music. You make bloofer troof 'ere. M8.
I digress…

Inb4sage42blepost
key fer
kek is with us
Ciaobabee

manage to formulate your posts coherently or don't post