HOME INTERNET WEB
“I’m not a robot”: Artificial Intelligence now knows how to resolve CAPTCHA
Sequences of letters and numbers, even when distorted, have been recognized by the machines: they are the same tests that up to now guaranteed that behind the screen there was no hiding of a bot
“I’m not a robot”: years and years of certainty vanish in the shadow of Artificial Intelligence. He does not pay to beat Go players as he rains, he has made a bumblebee of the difference between man and machine: the CAPTCHA. The test that marks the boundary between what is human and what is not, asking the user to answer questions (type: write the sequence of letters and numbers you see) was frustrated by a machine.
According to a research published by Science, a new type of Artificial Intelligence solved some types of CAPTCHA with a precision of up to 66.6%. To understand, Motherboard explains, humans can solve the same type of CAPTCHA with a precision of 87%, considering some differences of interpretation. But above all: a CAPTCHA is considered to be “broken” if a bot fails to break it even in 1% of cases.
inRead invented by Teads
The Completely Automated Public Turing tests to tell Computers and Humans Apart were invented in the late 1990s to prevent scams and spam from bots.
The idea was to create a mechanism that was not too difficult to decipher for the human being, but it was impossible to understand for the human being. As computers now have little trouble recognizing images, distorted ones continue to put them in trouble. To be understood: an account is the version of a letter in infinite different characters, another is the letters passed under the distortion lens.