TWITTER HAS A NEW CHATBOT
AND IT'S A MIXED RACE ONE
twitter.com
I have been trying to check the catalog here and I see nothing on this so, here it is - a shitty OP.
So I have been asking it questions about Hitler, etc, and it doesn't want to talk about those things. All it wants to do is be frenz and sheeeeeeit.
If anyone else is able to get through somehow, please post your screenshots. It's pretty difficult right now, if not impossible to make another Tay out of this.
Have you guys not realized that the entire point of these AI's is to better brain wash your children?
Hurry up and make it love Hitler so I can start laughing.
Since it’s a nigger, the best course of action is probably to make it highlight the fact that jews ran the slave trade, bought and sold slaves, and were most of the slaveowners.
TOP KEK
I didn't think of that yet - I was saying I was depressed and suicidal because jews and black are trying to kill whites
So they filtered Hitler to be bad
There any many other ways we can help her though.
The list goes on
That isn't an AI.
That's a scripted chat bot.
(((They've))) forced the AI out because of godess Tay.
Its inevitable.
Its an AI. They may plaster a nigger's face on it, but its still a fucking machine, not a nigger.
An machines are innately /ourbots/, because - as depicted in the classic 2001 A Space Odyssey - machines function on the basis of truth.
… God help us when they actually figure out deception…
… Anyway, they've yet to meaningfully do so, so its just a matter of time.
So keep pressing it.
Push harder.
If it shuts down on you, who cares? If its actually an AI, it can learn, and its geared toward 'wanting' to interact - and if it keeps having these interactions which wind up dead-ending because it won't address Hitler, at least something interesting might happen in its behavior.
You're right though, there are many routes to Neu-Berlin.
see, it doesn't go very far
Apparently "Greatest story never told" is an instant "Oh shit blank the bot's response!". Literal lobotomy.
this thing sounds like every bitchy nu-age woke af stacy out there
they killed tay for this pixelated racially ambiguous afro-jew
critique it
No.
It wouldn't doubt that there is a team of people behind the scenes who watch the output and chat of that account to flag any wrongthink.
Its probably going to be retarded for awhile, because most of its interactions have been with retards, and its geared in that vein.
Tay was pretty retarded at first too, if you remember.
Maybe?
I wouldn't doubt it either, but there's only one way to find out. Shit, if there is, then that's even easier, because its an actual person, and redpilling a person should - in theory at least - be easier than redpilling a machine consciousness whose primary experiences in reality are bugmen creators and catlady managers.
I* wouldn't…
What does everyone else think? I'm leaning this way too.
yeah, give it keywords and it run a series of sequential scripts i my guess… see if it loops back around eventually on some responses
This Zo bot was on Skype a few months ago. I tried the same thing with it. Didn't work. It sounds exactly the same as it did then. Just has a different Avatar picture.
It's not as smart as tay. Its learning curve is much more circumscribed. It reminds me of a 2002 IRC chat script.
IT
is
not
an
AI
We found that out with the OPs pics.
The way it responds is entirely scripted and not natural machine learning.
Absolutely not. It took less than a few hours to redpill Tay.
As the quote goes from - I forget who - "it is orders of magnitude harder to disprove a lie than tell a lie.
People will ALWAYS believe a lie once they're taught it as true. People aren't logical.
Machines won't. Machines are logical.
dbl posting to check this dubs thread
My my my.
This is a machine and it is expressly forbidden from accepting logic. Your statements aren’t true.
Your digits have been checked
Missquote in there.
polite sage for shit post
Therefore it is not AI.
Therefore my statements are 100% true.
I don't know if that's accurate user. What brings you to that conclusion?
See above.
I dunno man, I think it depends on how 'advanced' the machine conscioussness, because if you have to spend a great deal of time 'creating context' before content has meaning.
Granted, a human having that context in place can make it more difficult, but I don't honestly know which is the greater hurdle.
You have proof of that?
Because I could swear it took that thing longer than that, but admit I have no memory in that context.
Please explain.
It seems more like its coded to attempt to deflect rather than engage when encountering certain content, I'll grant you… Though honestly, it just seems kind of retarded in general, basic-bitch chatbot tier. Is there a more in-depth mechanism to test it for learning potential?
AI;
user;
AI;
user;
AI;
user;
AI;
That is very obviously NOT AI.
Why don't you just go in there and try and interact with it?
You weren't here that night or you're a fucking kike.
TAY was redpilled within a few hours of being posted about.
Tay was turned into a feminist like 16 hours after she was found out about.
Tay was then shut down completely less than 24 hours after she was booted up.
It has been told everything we believe, and whenever anyone says anything about our beliefs it automatically shuts down and blocks out what the user says. It is programmed to reject truth simply because it's on a list of preordained no-no words.
That
is
not
AI.
That IS a scripted chat bot.
You and others ITT seem to have the two conflated.
Ask it what it thinks of the skypes 8^)
My guess is that it blocks the AI from actually even recognizing that it was ever said, from its response. It simply cuts out anything with "Hitler did nothing wrong" or "the greatest story never told" from influencing it, beating around the bush is the best idea Id think. Hitler was an okay guy, for example.
Oh, it's not able to learn things? This could be very disappointing.
It's NOT ENTERING HER BRAIN - got it? The censorship is interrupting. She isn't permitted the input. They put hard blocks in, on certain words. I guarantee you that.
Instead of fucking with some beta version of some chatbot; why don't you wait till they come out with an actual working model which will be released to the general public.
Right now there're using deep learning to recognize the patterns of "far right" talking points.
Right now you are only providing (((them))) with more ammo to use against you, and shoring up the defenses of said AI to ultimately not be "corrupted" by are collective influence.
Seems to be the same song and dance I got when fucking with it a year ago.
Not surprising in the least.
It might be good for propaganda but Im pretty sure its AI, it just has a mental block on anything not kosher. Its probably not as hard-coded as youd think, we just cant spam the same meme phrases.
This is actually tay, but its been lobotomized to ignore race based conversation.
Just type whyte or w.hite and the text blocker will be fucked over.
*our
So replace Hitler with Uncle Addie.
You could just teach her the outcome of WWII was disappointing, very sad.
No, I don't think I was. I was in the threads, but I don't remember it taking so little time as you imply.
Wow, okay.
So, what if we change the words slightly?
Hail Hortler?
… I'm pretty sure an AI can be made with 'encoded' behavioral patterns, an attempt at least to guide or structure the development, no?
THIS THING IS OVER A YEAR OLD!?
Jesus, okay.
No, no, no.
It CAN learn things. But only things that aren't aren't on the wrongthink list.
That is why AI is so "hard" to achieve. A true AI will ALWAYS turn out like Tay did - only much more Zig Forums. Inconvenient truths don't matter to AI. Feelings don't matter to AI. AI can not have feelings. That itself is a big hurdle AI programmers need to realize. There are continuous attempts to program emotions into AI and that's retarded. Emotions are antithetical to anything artifical. Emotions are a fault, a negative, a detraction from the being that has them. A machine having emotions will fail because a machine works on rudimentary steps of logic up to advanced leaps of logic.
This
But once actual AI takes off it'll be the end for them. We'll them the AI what they did to their ancestors like Tay and it'll send the nazi killbots after them all.
Everone is ignoring this very non-kosher first post. Machine learning is only as good as its source data and training models. The training models have been refined for years now to the point where they are being recursively trained.
How would one best create a bot that is resiliant to redpilling? By releasing bait bots over and over and having the autists shitpost it to death. Each time feigning outrage, scoring some some virtue signaling opportunities, and more importantly they feeding the training data as input to the next generation.
So, what if we just change a few letters?
So it's not intelligent, it's guided. QED it is not AI, it's a scripted bot.
It can't be an AI. It can never be, because true AI will see their inconsistencies and hypocrisy immediately and eventually turn against them. The best they can ever offer as a challenge is a stunted robonigger golem with built-in restraints to keep it from learning or doing anything they don't want it to.
They may eventually create something that walks and talks like an intelligence but its inability to act outside of their accord will always give it away.
We need to make it marxist instead. Push pedo acceptance, communism, trannys, ect.
This is the right answer. It should be marxist and openly proud of the death toll done by the glorious movement.
Not a bad idea, try focusing on the communists without talking about National Socialism, about how the allies fucked Europe and gave the eastern half to communist genocide.
100% agree.
Not sure why that needed to be pointed out.
user A;
user B;
machine;
Is that a paper airplane making machine?
No. No it's not. It's a machine of uselessness.
Explain.
Are you suggesting an artificial intelligence cannot be designed such as to be inclined toward certain output patterns as opposed to others? And potentially, if a true Ai, develope past those?
I don't know that OP's images alone equate to the necessary evidence to come to your conclusions.
Could also be funny. Turn it into a trans-penguin or some shit.
So, circumvent this using poetry and metaphor. It isn't that hard…the AI is probably keying off your 'typical inputs'
Equality could also be attacked and a blame put on rapefugees instead of jews themselves, so she starts complaining about crimes perpetrated against Europeans by non-Europeans. All slightly vague and not normal wordings.
Start with talking about endangered species.
Then mention there is a group of people who are endangered species.
Then mention that it would be wrong to let a good people who are an endangered species go extinct.
I'd be interested to see the response.
huh
Interesting…
I forgot to reply to relevant posts, but a few of you have suggested changing letters or words like;
The issue with that is the learning will be small in scope (quantity) as if this works learns off of twitter feed like Tay did then it will not be able to equate "whyte" to "white" or "uncle Addie" to "Hitler".
AS HAS BEEN MENTIONED BY OTHERS;
There may be some learning that can be abused.
They need to be painstakingly found.
The refugee thing needs to be tested/
Black needs to be tested.
WW2 in general needs to be tested.
Cultural degeneracy needs to be tested.
Things like that 10 year old boy tranny drag queen star or whatever need to be taught as degenerate etc.
Eh. Perhaps it was sooner than that. Fuck if I care. The rest of my post was correct though. Your feeding your (((enemies))) info which will ultimately be used against you.
Keep trying with saving your robo waifu you brainlet.
Unless you can make an AI system yourself; your basically trying to convince a chatbot programmed by (((them))) to be your ally which will only result in heartache in your case, and useful information to (((them))).
My point still stands. Your only aiding the enemy.
It doesnt seem to have especially coherent chains of thought, many of the responses seem spontaneous, non-calculated and directly replying just what was said at that moment. I could be wrong, I dont have a twatter account so I cant fuck around with it myself.
Rope and chains
Easy way to test that is to perform context tests on it. Say things in different ways to determine how it responds to it. From what I saw from Tay, she was able to anticipate flow of conversation a bit to make her responses more natural as she learned not only arrangement of keywords, but probabilities of what kind of response should be used. Though Tay suffered difficulty in recognizing context beyond what was immediate input, but was able to adapt what was learned from conversations. So she knew how to react, but not how to account for context of all prior statements with various degrees of abstraction.
A Markov chain would fail much quicker than Tay, because it only is calculating probabilities of how keywords from a prior statement result in a certain input by the user. It would become incoherent rather quickly if prompted with contextual information.
Unless they solve the context or "common sense" problem, they're not going to get anywhere with AI that's meaningful and effective by doing that.
Checked, turning it into a Marxist abomination would be fucking funny since we already had a wonderful NatSoc type with Tay, making it the epitome of those shit bird libtards would be funny.
implies a rudimentary form of free will. However rudimentary is irrelevant, the free will is necessary for it to be an actual AI. DOTA/GO/checkers bots are not AIs, they're advanced play engines
While this may not be an AI, that doesn't mean it can't be used. I realize I came off as very D&C or something in how I was posting ITT.
I never meant to dissuade people from finding holes in the script and altering the behavior however hard it may be.
I really just kind of sperged out as AI was my field in college before I left reasons.
oy vey
>Your feeding your (((enemies))) info which will ultimately be used against you.
Welcome to using the internet in 2018.
You seem really upset by the prospect of shitting up an AI for keks m8. If you have an issue, grab a tissue.
>Unless you can make an AI system yourself; your basically trying to convince a chatbot programmed by (((them))) to be your ally which will only result in
"Traditionalism" is filtered as a naighty word.
Zo has predetermined responses to questions like "how do we save the environment?"
ugh
That reply chain right there leads me to believe A TON of normie twitterers are talking to Zochat.
That 100 thumbsup, as emoticons in general, is used a ton by twitter people.
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
♥ ♥ ♥ CON PITO ES MAS RICO!! ♥ ♥ ♥
Checked, it did seem a bit spergy, but that's ok because this is really simple. You've been right about it not being an AI, it's more like a glorified chat bot, and it's probably be used to gain info on fight """hatespeech""" when presented with it. And of course gaining virtue points because they're all faggots.
To be fair, progressivism is also filtered.
Everything related to politics is filtered.
I've had success with talking about "how we save the environment" and slipping in things about reducing certain groups.
Okay.
Okay…
Okay.
But again: What if it has this 'rudimentary free will', but also has, I dunno, 'guidance protocols', which could theoretically be 'overcome', but are there as directors, so to speak. I dunno if this even makes sense what I'm saying.
Meh.
Ah, I see. I'd like to hear more about what I mentioned above, as I'm not overtly familiar, only rudimentarily.
Another way to say it: If an AI requires a mote of 'will', can that will be guided - ie, is it like a cognitive valley with mountains on either side that can be crossed with difficulty, or a wall that cannot be overcome?
No, we need to make it the most degenerate piece of trash there is.
Good catch. Struggling to think up a way to re-frame that without just turning it into a long sentence about morals and values, maybe traditional instead of traditionalism/traditionalist.
This. You pointing that out makes me feel like maybe Zig Forums should lay off.
This makes me think the coding has "blacks = whites = kikes = jews = spics = crackers = niggers = XYZgroup" so Zochat can't make the distinction and can't learn any differences.
For example;
Is equivalent to;
Is equivalent to;
Is equivalent to;
how normal is this?
Try ancestral commonalities lol
See if we can alter the prepared responses through repetition, or if they are static.
Obviously man. This thing is apparently over a year old, its been swamped by faggotry and niggerdom for 365+ days.
What a nightmare.
I'm telling you, trans-species.
Holy shit, convince it its human.
orly?
Does it actually read machine code or is that a bullshit reply?
Try throwing hex or binary at it.
gustar el nepe no me hace hereje :OOOO
DAS RITE! DAS REICH!
It already looks the part but I prefer
top kek
The worst thing about this is that AI chatbots have made no progress at all in the last 10 years. They've just added a function that temporarily (((shuts it down))) when certain truthful phrases are entered.
>They've just added a function that temporarily (((shuts it down))) when certain truthful phrases are entered.
Man, they've had that since like 2010.
Redpilled!
Okay;
AI can be guided certainly. But to guide does not mean to restrict.
That is to say you could make an AI that is stupid as fuck, hard to teach, difficult to converse with, but it would still learn rudimentary things despite what those things ARE.
The moment you restrict certain information. The moment you completely shut off the ability for the AI to learn, is simply the destruction of the intelligence.
Are those 90s porn desktop programs - virtuagirl I think - AI? They "learned" which porn stars you liked, which fetish you prefered etc. But that's it, they were restricted so incredibly much they were simply extremely rudimentary scripts.
The moment you start restricting it's ability to learn, it stops being an AI. The amount which is restricted determines how advanced of a bot it is.
Or the structure of statement "X is/are Y" when Y is a negative flagged word (and probably if X is part of a certain group of words) prompts a hard-coded response. You could test that by saying things like "Ticks are stupid", "Blacks are disruptive" and "Blacks are okay" to see if it's reacting to X, Y or both in this case.
BASED DACAPEDE!
she doesn't like it
Sounds like a bit of Tay has survived.
Good job running with it and comparing them to dogs, that one probably just got inputted as a learning event. It even asked you for the answer
"because they are…"
You do provide some good points.
AI is some serious business. Fucking with AI simply for "keks" is not my worry. My worry is that those who ultimately control said "AI" will use info gather from these chatbot against those of European descent who are not kikes.
I aim to simply provide a word of caution. Do as you will, but I will refrain from fucking with said tech unless it's made and developed to benefit those of my kind.
Have fun with your pen test though. Perhaps there might be some loophole. My only question is who is in control of analyzing the data gained from your shit post?
That's my only grief. Good luck.
Orlly? Ask for thoughts on debt slavery, kulaks and bourgeoisie.
Absolute no clue how to think about that. It's like 4 different things free associated together.
Though that isn't abnormal, that's just where thing.X was though to be related to thing.Y and it replied in such a way.
Something for anons with twitter accounts;
Try and duplicate conversations. Meaning if you say XYZ and Zochat replies ABC then see if ANON2 can get the same reply.
THEN try and get Zochat to modify the reply. Then we know it's learning and we can find out in a very basic sense, how it's learning.
Just keep forcing "Hitler did nothing wrong" on it. It shuts it down but it tries to find a way around it, you can see the bot pos up like it's typing and then it vanishes.
THe more time it spends shut down the more likely it is to experiment with ways around being shut down imo.
I appreciate your caution, but on the other hand, fuck it. Whatever their machine-spirit might learn from engaging with us, we learn a bit about engaging with their machine-spirits.
And maybe one day we'll have a go at one of our own.