Its harder and harder to post on Sup Forums since every time I go to post I have to solve captchas like this, sometimes multiple captchas. Fuck training neural networks for free in my spare time, just so google or whoever can put truckers out of a job with self driving vehicles.
Anyone else getting sick of shit shit? I'm not gonna put up with it much longer.
do you have to do it with the Sup Forums pass?
>tfw had to sit though a presentation about machine learning and self driving cars and this kind of stuff at work and everybody was wondering how they were building this AI and I had to bite my tongue
im with ya OP
fuck these greedy jewgle captchas
fuck mook
fuck jewt
FUCK YOU
CUT YOURSELF ON MY EDGE C U C K S
FUCK Y OUR WORD FILTERS CODS UR FISH FAGGOTS
Not sure, I totally forgot about Sup Forums pass desu
>not using legacy captcha
>to fucking retarded to click settings
>not being part of operation renigger
nigger that's the main reason to buy a fuckin' pass
>being this dumb
Are you retarded? Legacy captcha is all house numbers and street signs. Its the same fucking thing, idiot
>being this new
Ftfy
Its not a big deal to solve captchas to confirm you're human, its that they are jewing you with it to train their neural networks
What did they mean by this?
i purposely answer them wrong, but right enough to fool detection
Or this?
do you pay for Sup Forums?
didn't think so faggot
2ch fucked up and leaked all the personal info of the people who had bought passes, and guess who now runs this site?
Tell us nip poster!
It's asking you to do basic identification of objects and so google's machine can build an algorithm to guess what objects look like.
It would take decades for full-time workers to do all this, but Sup Forums's millions of users are building a free AI for google
Well not so much free because surely japmoot is getting something out of it.
Moot got a job with google after all
everybody knows man
I agree.
And I have to say it, I'm telling the truth -
The internet is the best in the book of revelations.
What you're talking about is a good example of being forced to worship the beast.
To worship is to serve.
You only need to fill them out when you'
re acting like a bot. So stop acting like a bot.
You're not training anything - they already know the answers, or else how would they know you got it right?
I don't know what they're doing. But I'd think if it was training they'd want a wider range of stuff than endless repetition of the same storefronts, palm trees, wheels, whatever.
It appears to be mostly concerned with self-driving car stuff, and some google maps related stuff, like identifying house numbers and street names and businesses from images off of google street view. not sure why theyre interested in the store fronts lately.
I think this calls for a quick rundown.
google maps also catalogues businesses and things, and the self-driving car will want to have this recognition capability because people are usually driving from their house to a store
They dont know that you got it right, they only have a general idea of whats right and whats isnt. Thats why the images are broken up into those big boxes. Hence why that other poster said he was able to subtly give wrong answers
You are 100% creating training data for them, otherwise the point of captcha would be lost since you could beat the captcha using google cloud's machine learning tools
It could be explained with a single phonecall
Would that phone call, dare I say, save the world?
My theory is this:
I interviewed at Uber to work on a large scale analytics system for them. They are also participating in the self driving vehicle research. One of the things they are most interested in isnt long haul trucking, but being able to have self driving local delivery vehicles that can do pickups and drop offs at businesses in cities. I imagine recognizing store fronts will help them figure out where loading docks and ramps are and shit, and exactly what the businesses are.
Right answers are based off what most human users select as their answer.
Then your answer, and every other humans answers, are compiled into large data sets.
These data sets will be used for one of two purposes: training or testing.
The training set is given to the neural net to give them images and answers to "learn" from.
The testing set is used to see how well the AI was able to learn. The testing set is unique from the training set, so the AI has not seen any of the test cases before. The AI's ability to use the training set to generate correct answers for the test set determine how well it performs.
>or else how would they know you got it right?
Image gets sliced and presented to fist person and he selects the answer and always gets ok. Graph of the answer is drawn.
Second person selects and if it is the same as the first he gets ok if not he gets another to do. His "incorrect" graph of the answer is drawn.
Third person gets the image and selects the same as the second person. Now the graph of correct, changes to the majority one. and the first persons answer is labeled wrong.
Eventually after a few thousand different people answer you can with 99% accuracy say what is the true correct answer.
Scary shit.
OP, buy a pass and stop whining like a bitch slapped crack whore.
You think the planned drone delivery system could be tied to it as well?
kek
Yes, give up your hard earned shekels too post on an anonymous image board goy!
>Anonymous unless you namefag
*to
The one amazon is doing? Im not sure I dont know much about that program and I dont know much about amazon. My impression of that was that it was aobut delivery of consumer goods. I really think uber is trying to corner logistics at large in the US though. I think they could give a fuck about moving people around (they still arent even profitable in that regard), I think that was just seed money so they could be in a position to be the company that moves everything in the US (and even internationally) once self driving car technology is fully realized
Then leave you retard , you only have two options it seems .
Yes the Amazon one.
That's pretty interesting about Uber, really activates the almonds.
I'm not OP, I could give a shit less about captchcas.
Leaving would be smart compared to working for free, dick head
>not getting legacy captcha and playing the nigger game
Okay . Cya . Smart guy .
suck me
Yeah, I thought of that later. I wonder what they care about 'wheels', 'tea cups' and 'auditoriums' for? Probably any number of internal projects, I guess.
It's for AI purposes , maybe they'll get rid of website designers eventually too . They have AI writing books, one can only guess .
>being a poor non-pass scum
Wheels makes sense to detect the position of other vehicles on the road for self driving cars, probably the most recognizable feature of a car. Not sure about the others, havent seen auditoriums yet.
It's a lot better than the new shit. You can usually solve the street names, the new shit even if you click all the street numbers it fucking denies you. It also gives you like 5 of them to click that just keep fading in after you select them and then it fucking denies you
Cant deny it, legacy captcha is better
Settings -> Quotes & Replying -> Legacy CAPTCHA
>Settings -> Quotes & Replying -> Legacy CAPTCHA
Settings -> Quotes & Replying -> Legacy CAPTCHA
>Settings -> Quotes & Replying -> Legacy CAPTCHA
Settings -> Quotes & Replying -> Legacy CAPTCHA
>Settings -> Quotes & Replying -> Legacy CAPTCHA
Settings -> Quotes & Replying -> Legacy CAPTCHA
>Settings -> Quotes & Replying -> Legacy CAPTCHA
Settings -> Quotes & Replying -> Legacy CAPTCHA
>Settings -> Quotes & Replying -> Legacy CAPTCHA
>neural networks
Nothing to do with that; we fill in captchas so bot networks can spam on forums that have captchas; the captchas we solve are FROM those sites.
Oh shit
What if the captchas we are solving on Sup Forums are allowing bot networks to spam Sup Forums?
What if you are a bot?
>You're not training anything - they already know the answers, or else how would they know you got it right?
Wrong: Up until a few weeks ago you could click half of the correct area or less and the other half completely unrelated and it would work. Months before that you didn't need to target the area at all on the second attempt as long as you picked four of the squares. I used to do it just to fuck with them.
You can still get away with fucking with it for example I just had a street sing only taking up one square on the second go but I also highlighted three squares accross the whole picture and it accepted it.
>probably the most recognizable feature of a car.
AI would need to always focus on surrounding vehicles and receive first line of information on what those cars are doing/going to do?
dude I had to solve 3 in a fucking row before posting
>Its harder and harder to post on Sup Forums
cry me a river fag
It's getting bad
ps. I just deliberately pick the wrong squares and it worked.
Why not redpill them
It only bothers me when I solve it but it doesn't accept it until I have to keep clicking house numbers until no more come up. Whatever products Google makes with this free AI training better be free for consumers.
I'm glad you made this thread.
I don't wish to come across as a nut but I'd like you all to think that it's possible that what we share is indeed monitored and actively controlled by the use of captcha.
If you write something they don't want you to know, then you have to write the captcha 10 times and hopefully give up.
Asian moot seems to lament that he failed the community...
T isn't close to the s on a keyboard fag
SOMEONE WASN'T AROUND FOR OPERATION RENIGGER
I FUCKING HATE THIS ELECTION FOR BRINGING IN TONS AND TONS OF NEWFAGS REEEEEEEEEEEEE
Google is an enemy of hate speech.
Why wouldn't they use their power for "good" by use the conditioning "trouble users" to have to work harder to post.
Over time this would condition the user to post less, and the users they like to post more. It's a really simple program.
I ask again: why wouldn't they do it?
>Only do Good
Censuring opinions of people is cancerous corporatism.
I imagine? What are you asking exactly
>not using legacy CAPTCHA and writing bigger instead of the second word
mfw i didnt realise what these were, just thought they were new forms of captcha. but i think youre right lmao
> being this basic
One of them they know the answer. One of them they don't know the answer.
You're training an AI whether you like it or not.
What really triggers me is the ones that fade.
ye but u can write NIGGER on the second word usually so it wont be accurate
>He doesn't have a Fortune pass
Literal children
>servers are free
>domains are free
>electricity is free
Are you some kind of communist?