Have you ever tried to go to a website but you were stopped beforehand in order to prove you are not a robot?
Well, what is happening behind the scenes may surprise you. Here’s what really happens when you check the “I’m not a robot” box.
What is reCAPTCHA?
The “I’m not a robot” checkbox is a CAPTCHA, or a “Completely Automated Public Turing Test to Tell Computers and Humans Apart.”
Essentially any type of test, whether it’s selecting the image with stoplights or typing out a word that’s blurry, is a CAPTCHA, which is used to stop bots.
reCAPTCHA, which you might recognize from the “I’m not a robot” checkbox, is the specific CAPTCHA program from Google that is used by millions of sites.
How does reCAPTCHA work?
So how does the “I’m not a robot” CAPTCHA actually work?
It’s not actually the act of clicking “I’m not a robot” that confirms you’re not a bot. It’s what happens before you click it.
Related
“When you click on that checkbox, the site sends back a bunch of information to Google,” John Lloyd, chief technology officer of Casaba Security, told Reader’s Digest.
This information can be your cursor movement as you go to click that checkbox (apparently humans move their cursors with more randomness than a computer), your cookies and device history, according to Cloudflare.
Google uses that information to determine the probability that you are a human or a robot. Google then sends that score to the website, and if the score is high enough, you’ll be let into the site.
Simply put, by clicking “I’m not a robot,” you are giving Google permission to analyze your online behavior to determine if you are a human.
Can AI solve CAPTCHAs?
Ironically, it turns out robots may be better at proving that they aren’t robots than humans.
A July 2023 study by researchers at the at the University of California found that AI bots were more efficient at solving CAPTCHAs than humans.
By comparing the speed and accuracy of 1,400 participants versus AI bots in solving 14,000 CAPTCHAs, the researchers found that bots were far better than humans at beating these tests.
The bots were 99.8 accurate while humans ranged from 50% to 84% accuracy.
