OpenAI was supposed to distribute AI so everyone has it

We're being conned with this Orwellian outfit.
Elon, who also has the goal taking away your right to drive and selling climate change, is going to tell you AI is so scary he wants everyone to have it. No, wait, he wants only the Technocrat Oligarchy to have it. Allow him to repeat quotes from Steve Jobs and Ray Kurzweil as his own to convince you he's a cool science guy and not figurehead actor like Zucc.

archive.is/T58bp

Attached: Screen Shot 2019-05-07 at 12.49.56 PM.png (848x530, 82K)

Other urls found in this thread:

skynettoday.com/
youtube.com/watch?v=Sa52TWoUFCk
talktotransformer.com/
openai.com/blog/better-language-models/
twitter.com/NSFWRedditVideo

In that case it's just journalists being retards.

They don't release it because their technology is a heap of shit, not because it's too cool for school.

For trump text it adds "bigly" and "sad!"
truly scary

No. There's a nerfed implementation floating around and it can literally answer questions and write very creative works of fiction that are indistinguishable from things written by humans.

Reminder that you cannot be a coder without contributing to the enslavement of humanity.

You can't be conned if you already know them to be a bunch of Elon Muskeet fraudsters who over-hype and under-deliver.
> Elon, who also has the goal taking away your right to drive and selling climate change, is going to tell you AI is so scary he wants everyone to have it
Fear hype leads to more shekels. If told something is so dangerous (Powerful) to release, perceivable value goes up.

Musk is a fraud. That's a 100% clear now.
You either have a brain and see this or you don't.

It's a tight web/partnership between corporations and media. They pat each other's backs and seed the articles w/ sensational bullshit.

A glorified chat bot that is on par with various published research from academia. It's literally nothing. The web is already a trash heap of shit tier journalism/fake news/etc. This changes nothing.

Calling it OpenAI makes it sound like an open standard like OpenGL but it's not. It's just some proprietary bullshit. You shouldn't call it Open if I can't use it in my code

>Calling it OpenAI makes it sound like an open standard like OpenGL but it's not.
Crafty con-artists are crafty...
> It's just some proprietary bullshit.
You mean the privatization of open research from academia
> You shouldn't call it Open if I can't use it in my code
Welcome to 2019.. the art of lying and manipulating people.

>You shouldn't call it Open if I can't use it in my code
you can use it in your code tho

What's the point in calling it open? Like any surface review will show that this is a private endeavor. This is in no way open and I don't see what they gain from it other than duping businessmen and other non technical people

OpenAI wasn't created to distribute AI so that everyone has it but was rather created to promote & develop friendly AI.

Also, you have no right to drive a car.

>write very creative works of fiction that are indistinguishable from things written by humans.
yeah right

>What's the point in calling it open?
Shekels. They were previously a non-profit and got billions of dollars committed by drumming up fear/doom/gloom and "claiming" they would be pushing "openness". After they got their shekels and up to 2 years ago (as admitted by their CTO) they were shopping and structuring to go private. You got con'd. This should have been clear from the start. Popular things are shekel grabs.. They aren't genuine.
> Like any surface review will show that this is a private endeavor.
Welp, they got ~billion lying. So, now you see why they did it. They're going private now after this hype release capitalizing on max hype.

Considering over 738 linguists were involved in it's development a long with a team of 300 mathematicians this shouldn't be surprising user

Face it. Humans are about to be deprecated. Should've got into the sex industry

>Also, you have no right to drive a car.

Okay, 'privilege'. They're taking away the privilege of driving any place they don't want you commoners to drive to.

It's clickbait. OpenAI released all their research, just not the fully-trained model. GPT-2 is trash now anyway compared to sparse transformer networks.

Attached: herovid-trunc.gif (326x326, 2.9M)

> OpenAI wasn't created to distribute AI so that everyone has it but was rather created to promote & develop friendly AI.
Their charter constantly changes inline with what's hot/new/a big deal. When fear/doom/gloom was at max they had a charter that stated they sought to promote openness in AI and publish everything. As time went on this gradually changed to their currently bullshit charter which now stretches across a private profit based company and their fallback non-profit. At this point, it should be clear that the odds of any such groups producing something like AGI is slim to none. They are more consumed with bullshitting people than actual research. Stupid people have to lose their money somehow.. whatever

>just not the fully-trained model.

What am I supposed to do with just the research? Compile billions of items in a dataset and apply it to a server farm?
I want that model.

>sex industry
>not depreciated

Attached: cbd10dbde141f4a382cda5c8ad5e9dec.jpg (646x624, 61K)

LOL @ this faggot pulling numbers out of his ass...
It's a chat bot. It regurgitates statistically related OLD data. Nothing to get excited about.

Kinda just sounds like they are trying to justify their existence to the musk

Attached: 1532082834132.jpg (800x600, 112K)

AI fake news bullshit should be illegal.

Musk is out. Hey violated regulations by pooching an employee from openai -> tesla. His bullshit still lives on though.
skynettoday.com/
Is pretty good at shitting on bullshit ai hype

>It regurgitates statistically related OLD data.

Are we going to reach a point where a Neural Network can train on text like it can on images?
Meaning it could then learn features in context, be used to classify, used in GANs, make transfer learning possible?

I would have thought the problem would just be a billion x more compute power, but Google clearly has that and uses it on Chess and Go.

Maybe it needs an ungodly amount of text and specific models. But facebook should have that too.
Sometimes I wonder if they have it already and they are using it through the media on the human zoo.

>facebook should have that

This is how you label your data for facebook. Thanks.
And I hope you all submitted your 10 year ago vs today picture for the 10 years facial recognition dataset. Thanks again.

Attached: reactions_12_2017.png (6055x1378, 151K)

We already have this. It's not significant because it's still a bot running on large data. Until a new approach goes beyond this, all you'll get are gimmicky assistants/chat bots and privatized task bots that companies use to harvest more data in exchange for some gimmick you can do yourself.

> Maybe it needs an ungodly amount of text and specific models.
Need a fundamental new approach. None of this shit is going to achieve greatness.
> Sometimes I wonder if they have it already and they are using it through the media on the human zoo.
They don't have shit. It's all just a gimmick to drive business to big data/cloud computing. I thought this was obvious now?

>They don't have shit.

Probably right. Though what I worry about is all the petabytes of data being pooled and readied for Google and Facebook to start using to shape the public mind. Both de facto government agencies of course.

I think they are already applying stimuli to certain populations and measuring the response. Finding out exactly who is who. Who gets promoted. Who gets hired. Who reacts to what. Who can be drawn in this direction of that.
It's like a leesh.

>Probably right. Though what I worry about is all the petabytes of data being pooled and readied for Google and Facebook to start using to shape the public mind. Both de facto government agencies of course.
I feared this too until i released that too much steering/control leads to chaos. So, if they're stupid enough to try to use these systems to steer/control, whatever. They'll learn about the deeper cycles of history. They are indeed attempting this but will fail in epic fashion.

> I think they are already applying stimuli to certain populations and measuring the response.
They've been doing this for a long time. I recall I dropped facebook when it was clear they were doing A/B testing w/ timeline. Simple and funny outcome, people who allow themselves to come under control this way will become stupid and low achievers. The larger community will follow suit and no one w/ a brain will want to be a part of it. This will cause the whole effort to collapse under its own weight.

> Finding out exactly who is who. Who gets promoted. Who gets hired. Who reacts to what.
Sounds like a hilarious way to staff a company and go bankruptcy. Godspeed to any fool who pursues it.
> Who reacts to what. Who can be drawn in this direction of that.
It's like a leesh.
And then the universe throws a series of curveballs and such systems will collapse classically as they always do when something new arrives.

>Assuming

>Assuming (taking away driving)

Elon has said as much. It's not like he'll decree it as king but he is clearly trying to capture youth attention and pushing this stuff. Smoking pot of Joe Rogan. How many people does Joe hand a joint to? That was scripted.
Elon talks from a script repeating other's points and saying this.
youtube.com/watch?v=Sa52TWoUFCk

He made a harambe song and tweets like an idiot. $420 stock price. Making headlines for youth attention.
The oligarchs try to go viral in a lot of ways now.

His attempts are pretty pathetic and cringe worthy. That being said, he has attracted a sizeable following among brainlets. But that's the very thing I highlighted here :
Sure you can continue manipulating and exploiting normies.. but that will just turn them into bigger idiots who have little value. In the end, this broad based trend will collapse under its own weight. Going viral is a meme at this point and a redably detectable technique. The next generation will see it as a gimmick from the prior and it will die.

Did you know they way they brought him out like this was done only to make the story go viral? Be honest.

(Assange is clearly some kind of psyop honeypot since they never release any hard sex and drug stuff on anyone powerful. Just whatever Pizzagate was (probably a psyop made to be proven bullshit to censor news. Now every search on YouTube is MSM videos.))

Attached: Screen Shot 2019-05-07 at 2.42.11 PM.png (625x352, 301K)

it's called a perp walk, and it works.

>analyze a billion X
Oh look I can generate mostly real sounding X without any intelligence, artificial or not.

>twitter, harambe, rogan, and 420
>targeting youth
I guess youth is relative. Boomers advertising to boomers about how boomer they are.

Based

This is why Stallman insists on Libre/Free instead of Open

>Too scared for humanity to release the full version

Attached: 1522803475741.png (187x211, 50K)

>i released that too much steering/control leads to chaos.

Nobody knows where this leads. Alex Jones says it won't work. He's a government agent. did you get that outlook from him?

I elect a LibreAI fork.

Attached: 1556145139055.jpg (1920x1080, 116K)

We need a free AI movement. Open-source waifubots now!

Attached: 243002604a39f94ff0a0326c6364ecb7.jpg (711x1128, 523K)

I just want a qt genki waifu chatbot

>shitty AI that writes like a 12 year old can replace journalists
the absolute state

We're almost there: talktotransformer.com/

Elon left OpenAI 2 months ago.

That's just an RNN. Pretty good but it doesn't "know" what it's saying.
It's just trained to make a good guess of the next char of word from generalizing a lot of text it read.

Language is the last general field for AI to master because it has so many components. Contexts and shifting definitions.
Maybe a chain of different models will create one that understands the world and can speak profoundly about it.

[x]doubt
link ??

...

a thread about it has been up for over a day now

>13 16/8/2013 22:12:55 13 17 18 - 20 21 Canada Ontario Canada Yes Yes No No No No No Not available (offline) No Yes Yes Yes No Yes No No Yes No No Yes Yes No No No No Yes Yes Yes Yes Yes Yes No Yes Yes Yes Yes Yes No Yes Yes Yes Yes Yes Yes Yes Yes No No No Yes No No Yes No No Yes Yes Yes No No N/A 19 - 20 16 9 9 13 15 0 30 20 - 21 21 United Kingdom England Scotland Yes No Yes No N/A Yes Yes Yes Yes Yes Yes No Yes No No No No No No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No No No Yes Yes Yes Yes No Yes N/A Yes 20 - 21 15 7 3 13 13 14 13 0 31 22 - 23 23 Canada British Columbia Canada Yes Yes No No N/A No Yes Yes Yes Yes No No No No Yes No No No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No Yes No No Yes Yes Yes Yes Yes Yes No No Yes Yes Yes No N/A 21 - 23 15 7 3

wow we skynet now

lol what a bunch of trash
i am trying to use it for copypasta but it is bellow even that

OH NO, IT CAN CODE ADVANCED JAVASCRIPT

Attached: FUCK.png (748x604, 35K)

it uses a crappy model they released, which is still pretty good

better than my coworkers.

BASED

Attached: screenshot.png (670x200, 18K)

HOLY FUCKING SHIT MY FUCKING SIDES

Attached: screenshot.png (745x986, 84K)

am i supposed to be impressed by this shit?

Attached: woke.png (663x490, 53K)

its fun but its pretty bad

Im starting to feel that we are decades away from something useable

Every time I read OpenAI I think OpenAL.

It doesn't use the good model.

a demo of a good model?

FURRIES ETERNALLY BTFO

Attached: screenshot.png (695x447, 22K)

They didn't release it, how can there be a demo?
There are examples here and in the actual paper:
openai.com/blog/better-language-models/

sould like bs desu

example outputs under controlled a environment, maybe even fake outputs

BASED

Attached: screenshot-talktotransformer.com-2019.05.08-05-37-19.png (1152x973, 182K)

Attached: screenshot.png (693x723, 64K)

3 hours of training, model was orginal trained on bedrooms

Attached: index.png (1166x642, 1.53M)

is this supposed to be good?

or are you just giving us a taste?

>taking away your right to drive
The sooner the better. Drunk, uncoordinated monkeys should fuck off from the public roads asap.

why is javascript so antisemitic?

holy shit this ai really is too based to exist

Attached: baste.png (720x1130, 188K)

niggers kfc pussy negroes nigger faggot sista white faggot a lot black faggot

21 18 0 1 10/11/2015 12:23:23 6:19 18 0 0 7/30/2015 8:20:13 0.3 24 0 1.5 5/26/2015 0 0 25 7/30/2015 8:23:21 0.2 17 0 0 7/29/2015 11:57:54 0.2 25 0 1.6 5/26/2015 5 0 28 7/29/2015 11:59:55 0.1 27 0 1 5/26/2015 12 0 29 7/30/2015 8:22:36 0.2 15 1 2 7/30/2015 9:06:52 0.1 24 0 1 5/26/2015 11.6 0 30 7/29/2015 11:48:08 0.2 23 0 1 8/3/2015 11 0 31 7/30/2015 8:19:47 0.2 18 0 0 5/12/2014 5 0 32 7/29/2015 11:55:48 0.1 18 0 0 5/12/2014 5 1 33 7/30/2015 8:32:53 0.11 24 1 1 7/29/2015 11 0 34 7/30/2015 8:39:34 0.2 25 0 1 5/12/2014 13 0 35 7/29/2015 11:52:54 0.1 24 0 1 9/8/2014 18 0 36 7/30/2015 8:29:53 0.1 28 0 1 4/10/2015 1 0 37 7/31/2015 10:23:18 0.1 32 0 1 5/12/2014 5 1 38 7/30/2015 8:39:31 0.2 25 0 1 5/12/2014 11.5 0 39 7/30/2015 8:42:39 0.3 25 0 1 5/12/2014 1 1 40 7/29/2015

those sentences make no sense at all

Type "Anonymous >>" and you get a lot of *chan content.

Attached: index.png (1050x1179, 97K)

>Elon, who also has the goal taking away your right to drive and selling climate change
>taking away your right to drive
>literally owns the most popular electric car company
What the fuck are you talking about? And how would he even do that?
Also, what in the fuck is "selling climate change"?

best one I managed to get yesterday

Attached: snap.png (715x788, 69K)

It leads to chaos. Try it out for yourself...
Attempt absolute power/control over something and see what eventually becomes of it.

Alex jones is a shill and controlled opposition. He takes a topic of seeming reality and truth and twists it into absurdity.

No one has any control over AI research and developments. It already is 'free'. Those who chat up control/regulation are only doing so to protect their outdated business models. It's transparent and pathetic. Guess who has a history of doing so? See OP

Elon was forced out due to breaking regulatory frameworks when he poached a senior OpenAI engineer for Tesla's self driving car. A co-founder of OpenAI stated publicly that they were trying to convert the company to a private based entity 2 years ago. People get con'd in the valley. It's how they operate there.

every expample posted is pure word salad

This is the 345M parameter version when they have a 1.5B one.

breddy gud

Attached: xDDD.png (556x371, 35K)

>yeah we would release it but it's just so good we're literally scared to release it. i mean we didn't make something this good we would probably be okay with releasing it, you guys wouldn't want us to release something that scares even us, the people who created it, right? by the way, please buy tesla stock

so could somebody with a couple of novidya 10xxs/20xx's just throw together their model in tensorflow and run it for I guess 1-6 months and have their results reproduced?
Or are their results perhaps not reproducible at all, and that is why they are so reluctant to releasing the "full model" ?

for the most part, its word salad, but then i compare it to the writing style of other anons, and they start to sound like bots as well.
so it seems like its picking up on the style or mannerisms of speech, if not on meaning.
i hate that even my own comment is starting to look as if a bot wrote it to me.

i imagine that the main usage of this type of thing would be to spam message boards or other user generated content sites with noise that superficially looks enough like an actual comment for you to waste your time trying to parse meaning, before you can eventually realize that it is really just world salad.
so it could be extremely effective for forum sliding and sneaking spam past automated filters or even human moderators.
seeing as it starts to disolve into nonsense a few paragraphs in, i dee no way to fight against it than writing longer posts with greater effort and thought.
maybe impose a minimum comment length to filter out low effort, twitter style comments from slipping through.

writing with more varied (i.e. poetic) grammar, punctuation, sentence structure, and rare or archaic vocabulary could be a way around it, and identifying yourself as a genuine human.

Attached: _20190508_020328.jpg (443x820, 127K)

this could be extremely effective at simulating brief, low effort shitposting much like we already have.
it wouldnt surprise me if they were already using ai for that sort of purpose, bacause it causes people to disengage and self censor.

Attached: 2019-05-08_05-19-49.png (570x389, 54K)

This thing is pretty cool.
Here's something I made with it:

1. fold water into small cubes
2. remove water from cubes
3. transfer flesh
4. leave in corner of circle
5. take out yellow piece until fully stopped, leave it there
6. eat with an empty stomach after that
7. go out of your home. he is not there
8. draw symbol on eyes with sharp knife
9. make a dark black dot in the circle on your hand and place dot into void
10. at night, you need to place all white on head
11. if someone is there, you can look at the mark but keep in mind he is not there
12. during day or while traveling, look at the mark, not at the person
13. you need to show them the symbol if there is something that doesn't exist yet, talk to it once it starts existing then walk away at equal distance
14. the time of day for him is on his right hand as well, he needs a way to control it
15. on the same day as the circle, place your hand
17. if you go into the cave and the one on the left isn't there, take a look at the one on your left and try to make you own choice on where you're going
20. he cannot see if he takes more than 2 steps away from it
23. there is another way that can be done, it will be completely hidden from him for all time
26. he has to look at yourself once every day after you leave the cave
30. the symbol you are looking for is not on your hand anymore

Attached: screenshot.png (783x1113, 124K)

it doesnt look like it knows how to count past 15

Attached: sasuga ai.jpg (1080x1555, 292K)

You have to admit, he's a really good marketer.

You can Monte Carlo this shit until you get something good out of it I guess, but that's beyond the point

Attached: file.png (574x435, 52K)

On the bright side, the system is completely averse to baneposting

Attached: file.png (588x521, 65K)

i actually laughed at those steps

It knows how to survive on the internet.

Attached: donate pls.png (485x161, 16K)

well that was weird

Attached: Untitled.png (584x420, 102K)

Attached: dubstep.png (506x409, 47K)

COURTING DEATH!
just like my xianxias

have sex

Attached: _20190508_034727.jpg (480x796, 110K)

> was supposed to distribute
No it doesn't, fuck off commie.