Yo listen up here’s a story
About a little girl that lives in a furry world
And all day and all night and everything she sees
Is just furry like her inside and outside
Yo listen up here’s a story
About a little girl that lives in a furry world
And all day and all night and everything she sees
Is just furry like her inside and outside
Another fun fact: most captchas, including recaptcha, are used to train AI. Some of the puzzle it will know the solution to (which is what it uses to verify that you’re not a bot), and the rest it won’t know the solution to and uses it to train AI. If you train your eye a bit you can eventually make pretty good guesses as to which part is which, and then go on to give the AI training part of the captcha wrong answers. There’s something satisfying to knowingly giving the captcha a wrong answer and get the green check mark anyway, pointing a middle finger at the people who thought they could get free AI training from you.
It used to be used to train text recognition software. It was super easy to tell which word was known vs unknown. The known word just looked like normal text that was edited to be kinda “wobbly” while the unknown text looked like an actual image or scan of a page. I’d always put “ass” for the unknown word.
The current iteration of captcha isn’t as straightforward in terms of identifying which is which.
I hear what you are saying, and in today’s climate I probably wouldn’t strongly disagree… But I would much rather there be a crowd-sourced exchange of services training AI instead of completely ripping off work done by real people without any permission.
But that second part is what’s happening. You aren’t getting any compensation for contributing to the training (and no, merely accessing the site doesn’t count), and that data is kept proprietary by Google, not released to the public domain.