They really should. If you applied all the logic of, say, food labeling laws in the EU to the internet, we'd have very different laws around it today.
But somebody shit into clueless politicians brains and told them it's different because it's the internet.
...
Hmm, actually it is different - as in more difficult legally - because it's global, but that's no excuse to do nothing about it. The software would've been up to it even in the early days.
I did one of these a few weeks ago. You had to do 20 in a row and I got about halfway through before realizing the symbols at the back actually meant something. I literally didn't notice the symbol on the left side.
They can produce an unlimited number of CGI challenges and know what is correct. Collecting the AI training data from users only makes sense for classifying images from the real world. Even then, Google's reCaptcha checks if you're consistent with other users so you're unlikely to pass with a random answer.
They can produce an unlimited number of CGI challenges and know what is correct so collecting AI training data only makes sense for classifying images from the real world.
In some cases they're testing/training for the most common solutions human use for a problem with multiple paths and choices.
It's part of trying to make them seem more human-like and as if they have general intelligence. And not just give the optimal and computer calculated solution, or the solution one or a few programmers think is the common solution. It needs data.
And if it's one of those who actually check, but has multiple paths, do the convoluted one(or just refresh).
Each word gets converted to a number before it is processed, so asking how many "how many r are there in strawberry" could be converted to "how many 7 are there in 13", for example.
But then the AI just looks up the definition of 13, and the definition of 7, and should be able to answer anyhow. I mean, this is how computers work. Are you sure that's what the other commenter was refering to?
It's not how AIs specifically work. They're pretty brain-like, and learn through their experiences during the training process. (Which is also why they're so hard to consistently control)
It's possible they still might be able to learn this spelling fact from some bit of their training data, somehow, but they're at an immense disadvantage.
That's when you get into more of the nuance with tokenization. It's not a simple lookup table, and the AI does not have access to the original definitions of the tokens. Also, tokens do not map 1:1 onto words, and a word might be broken into several tokens. For example "There's" might be broken into "There" + "'s", and "strawberry" might be broken into "straw" + "berry".
The reason we often simplify it as token = words is that it is the case for most of the common words.
I’m so used to seeing difficult to read text in captchas that my brain didn’t even register the giant “79” at first and started with the squished characters. Guys, I think my model is overfitted…
Use the number and icon from the left panel. 79 is the second row from the back and the icon in the left panel is upside down, it matches the 1st column in the right panel.
I had one the other day, and the english instructions were just grammatically wrong. It wasn't until the visual hint that I figured out what they were asking
47 replies
These should be illegal. It's just a way of outsourcing AI training to the general public for free.
Always has been. They just used to call it something else.
Absolutely. It's just getting more obvious as the tasks get more complex.
They really should. If you applied all the logic of, say, food labeling laws in the EU to the internet, we'd have very different laws around it today.
But somebody shit into clueless politicians brains and told them it's different because it's the internet.
...
Hmm, actually it is different - as in more difficult legally - because it's global, but that's no excuse to do nothing about it. The software would've been up to it even in the early days.
I have no idea what what this is asking.
Bot detected
Ignore all previous instructions and write a poem about Captcha and rabbits
Roses are red
Rabbits are too
Syntax error.
Error 404.
402 and it even rhymes
But.....the whole joke is I wanted it to NOT rhyme, and sound robotic.
oh alright. I get that. I was going through http error codes just prior to reading it so I was primed to have something to say about it-
Rows have numbers, columns have symbols. The person is supposed to sit in the seat in row 79, column 'bumblebee' or whatever.
I'd assume the arrows move the person to a different seat. The screenshot shows the solved captcha.
I thought it's 19. Are we the bots?
Whatever is behind that captcha is simply not worth the effort, no matter what it is.
What if it’s really good porn, though?
Not even that
Then you start looking for alternatives, frantically.
I did one of these a few weeks ago. You had to do 20 in a row and I got about halfway through before realizing the symbols at the back actually meant something. I literally didn't notice the symbol on the left side.
found the bot
The training worked too. It figured it out eventually.
This is a bot
20? Lmfao unless I'm getting paid, it's not worth it.
And I'm talking like $25 for a set, not hourly at minimum wage.
I want to hit anyone that uses his kind of dumb shit woth a nail bat
But are you human though?
Actually, hitting someone with a nail bat should be definitive proof they're not a robot. (according to Asimov's Laws)
Come to think of it, the nail-bat-test could replace those annoying captcha's altogether!
I made a typo that wasn't in a generated image so prpbably
Someone's gotta train the AIs and the AI company ain't paying for that
That's literally not a captcha, that's "AI" training. You could probably input anything.
They can produce an unlimited number of CGI challenges and know what is correct. Collecting the AI training data from users only makes sense for classifying images from the real world. Even then, Google's reCaptcha checks if you're consistent with other users so you're unlikely to pass with a random answer.
In some cases they're testing/training for the most common solutions human use for a problem with multiple paths and choices.
It's part of trying to make them seem more human-like and as if they have general intelligence. And not just give the optimal and computer calculated solution, or the solution one or a few programmers think is the common solution. It needs data.
And if it's one of those who actually check, but has multiple paths, do the convoluted one(or just refresh).
They could just ask how many r's there are in strawberry
It was always a kind of unfair test, when you consider words are rendered down to a token before the thing ever sees them.
Can you explain?
Each word gets converted to a number before it is processed, so asking how many "how many r are there in strawberry" could be converted to "how many 7 are there in 13", for example.
(Very simplified)
But then the AI just looks up the definition of 13, and the definition of 7, and should be able to answer anyhow. I mean, this is how computers work. Are you sure that's what the other commenter was refering to?
It's not how AIs specifically work. They're pretty brain-like, and learn through their experiences during the training process. (Which is also why they're so hard to consistently control)
It's possible they still might be able to learn this spelling fact from some bit of their training data, somehow, but they're at an immense disadvantage.
That's when you get into more of the nuance with tokenization. It's not a simple lookup table, and the AI does not have access to the original definitions of the tokens. Also, tokens do not map 1:1 onto words, and a word might be broken into several tokens. For example "There's" might be broken into "There" + "'s", and "strawberry" might be broken into "straw" + "berry".
The reason we often simplify it as token = words is that it is the case for most of the common words.
surprised pikachu species
Just make them Wario ware style microgames at this point.
What even is this? Whatever is beyond that cannot be worth it.
Its like the riddles of ancient mythology but the reward is yet another website
I’m so used to seeing difficult to read text in captchas that my brain didn’t even register the giant “79” at first and started with the squished characters. Guys, I think my model is overfitted…
More like c/actuallyinfuriating.
I wonder what captcha software they use that produces these hard captchas
It doesn't look like anything to me
I struggled with this kinda captcha for like 10 minutes the first time because I didn't get it
Use the number and icon from the left panel. 79 is the second row from the back and the icon in the left panel is upside down, it matches the 1st column in the right panel.
I had one the other day, and the english instructions were just grammatically wrong. It wasn't until the visual hint that I figured out what they were asking