What is CAPTCHA? How Did It Develop?

Today, every internet user takes a test that they must successfully complete while performing various operations on almost every site they visit, even when entering the site: CAPTCHA. Developed by the team at the Carnegie Mellon School of Computer Science, this test is used to determine if the visitor to the site is a real person.

CAPTCHA has come up with many different themes so far. Today, he either makes us choose traffic lights or asks us to choose pictures that contain a certain object from among several pictures. Well, the one that makes us fail from time to time and have a nervous breakdown, “Am I a robot?CAPTCHA, which caused us to think for what purpose and how was it developed and up to the present How did it evolve? Let’s get to know this test that we encounter every day.

• Note: I recommend you to read the entire article, as technology is constantly being improved. Otherwise You may miss links to important points.

First of all, what is CAPTCHA?

CAPTCHA, “Completely Automated Public Turing test to tell Computers and Hhope Apart (Fully Automated General Turing test to Separate Computers and Humans). questioner, CAPTCHA’s algorithm it is itself. We need to prove to our interrogator that we are not a computer in our answers to the CAPTCHA. However, if we answer correctly, we pass the Turing test and have access to the content we want.

How did CAPTCHA come about?

CAPTCHA at Carnegie Mellon School of Computer Science Luis von Ahn Developed by Manuel Blum, Nicholas J. Hopper, and John Langford with accompanying directors. If the idea to improve the system thanks to a giant company appeared. Luis von Ahn told the story of the development of the system with his own mouth.

In 2000, still in the first year of his doctorate, Luis was one of the giant companies of that time. Speech by Yahoo’s chief engineer joined. The topic of the conversation is that Yahoo somehow 10 problems he couldn’t solve. One of these problems sparked Luis and his team’s idea of ​​CAPTCHA, which has made us shed a sweat today and has opened up discussions on artificial intelligence over the years.

Yahoo at that time programmer to open millions of email addresses He was trying to deal with his users and he couldn’t get past it. The programs that were written were able to create an e-mail address on Yahoo simply by filling out the form, and could do this endlessly. The huge Yahoo could not get ahead of this.

Luis von Ahn and his team also came up with a very logical solution for this: A test to distinguish between computers and humans. This test is for all ages and all over the world, in short. had to be resolved by any human, but computers shouldn’t have passed this test. That was actually the hard part. But for this, the team appealed to human nature: We can easily manipulate characters and text. We were able to recognize the language regardless of the form and environment they were in. But computers could only recognize these characters as shown to them. Of course, this was true for computers of that time. We will talk about this issue shortly.

This is how the first CAPTCHA system emerged:

The first version of the tests shown to the user in the CAPTCHA, unchanged consisting of two words and shown distorted it was a series. The computer was given the correct answer to this question and the user was asked for that answer. While humans could recognize distorted characters, computers could not recognize these characters automatically, so they could not complete the test.

Yahoo started using this initial version of CAPTCHA on the email compose page where it had problems. In the first weeks that CAPTCHA was added, the system was used millions of times and indeed succeeded in separating humans from computers. But with this system, a big problem started to emerge: Artificial intelligence develops with the answers given, and even CAPTCHA answers were falling on the black market.

CAPTCHA made the computer smart while separating the computer and the human:

These answers, along with users’ answers to the CAPTCHA, were actually logged somewhere. Thus, when a computer encountered a CAPTCHA with a known answer, it passed the test successfully by pasting that answer and was identified as ‘human’ by the computer. In fact, this situation became so serious that ‘CAPTCHA farms’ were established and people working with moneyby solving thousands of CAPTCHA, these tests saved correct answers for bots. Thus, CAPTCHA was forced to upgrade to the second version.

Second version CAPTCHA: reCAPTCHA:

The next development of CAPTCHA took place in 2005. This time CAPTCHA, ‘reCAPTCHA‘ was named. In this version, two words are no longer literal, of which someone has been made replaceable. One of the words was a word that the computer knew the answer to, and the other was a random word from books, articles, or any other content with irregular characters. The computer did not know the answer to that word. So how did the computer know if the answer to that second word was correct?

reCAPTCHA is the same two words, one known to the computer and the other unknown, not to a single user, was showing to multiple users. So multiple users ask the same question the answer was perceived as correct because it gave the same answer. During this process, thousands of articles and books containing bad spelling were manually scanned and the second word was formed.

Google’s acquisition of reCAPTCHA made the system even smarter:

By 2009, Google acquired reCAPTCHA. One of the first things the company did was to create the second word. what was needed was to automate the scan. The words with the gray background, which looked like they were taken from that old book we saw, were indeed taken from old books. With this method of the company, the computer scanning the texts became able to recognize corrupted forms of words. Computers no longer had trouble recognizing a garbled word when displayed. Computers have learned to read garbled texts.

A test conducted by Google in the field of machine learning in 2014 also revealed horribly how smart computers are. According to the results of this test, people can’t miss out on misspelled words. 33% accurate could read. Artificial intelligence, which once could not detect even extended characters, was excluded from this test. 99.8% got the correct result. Thus, in the CAPTCHA method, to a revolutionary change gone.

The second version of the second version: reCAPTCHA V2 and No CAPTCHA reCAPTCHA

After testing in the same year reCAPTCHA V2 version has been upgraded. In this version, users no longer encounter texts, like today. with pictures was meeting. To users select images with a specified object it was said. However, GoogleNo CAPTCHA reCAPTCHA (ReCAPTCHA Non-CAPTCHA)” has also put into effect a system called.

The first names to use the system at that time were Snapchat, WordPress and HumbleBundle. With the new system, users can “I’m not a robot” began to see the control box. As for the way the system works, it actually managed to separate users from computers to a large extent before it came to the test.

An always-behind system called “Advanced Risk Analysis” was monitoring users’ interaction with the site. Yes, we are constantly monitored on the Internet. It even registers how we navigate the page, improving technologies. If the user’s behavior matches the behavior expected from a real human, you can click I’m not a robot. you don’t get any testIt was confirmed that you were human. But interaction does not match the interactionthat is, if the computer thinks that a computer has entered the site, they will It was showing CAPTCHA. These CAPTCHAs were also the picture CAPTCHAs that came with reCAPTCHA V2.

Picture verification of reCAPTCHA V2 was of course also improving artificial intelligence. But this time your answers are mostly not by pirates, used by google. Even today, Google uses our responses to CAPTCHAs to distinguish objects within an image. So, for example, a self-driving car can cross a tree, a fire hydrant, a traffic light, or a crosswalk. can detect. As you can see, this system, which enables us to use the internet safely, enables Google to develop further. Win-win situation.

Today’s reCAPTCHA: reCAPTCHA V3

Google completely retired the first version of reCAPTCHA that showed posts on April 4, 2018. Today, many websites still able to continue using reCAPTCHA V2. However, in addition to this physically available test, reCAPTCHA V3, which was released on October 29, 2018, is also used. reCAPTCHA V3, taking reCAPTCHA V2 become more advanced has brought.

Websites using V3 instead of reCAPTCHA V2 since 2018, does not require any interaction by the user. These sites do not show the CAPTCHA test. On the contrary, informs the owners of the websites in case of a risky traffic. The user experience is not degraded in any way by CAPTCHA. Elimination of risks is left entirely to website owners.

It is useful to explain this working logic with the example of Google. You own a shopping site and have recently seen a high amount of traffic to your site. But this traffic is not converted into conversions in any way. So you think they are not real customers. At this point, all you know is that your site is getting high traffic. However, this traffic You don’t know what it’s aiming at, what it’s doing on your site.

That’s exactly what reCAPTCHA V3 lets you know. The reCAPTCHA V3 system, which you add to each page separately, tracks every entry made to the site. For example, the system you add to your product reviews page shows you that bots have entered this page to comment. Each transaction is scored by reCAPTCHA, and the site owner can distinguish between real comments and bot comments. Distinguishing bots is the interaction on the site used in No CAPTCHA reCAPTCHA. by comparison with real human interaction. is being done.

All this development shows how much the technology, which emerged as a solution to the problem Yahoo is experiencing, can grow today and How can you contribute to artificial intelligence? it showed. At the same time, it helped develop new technologies in order to track our fingerprints on the Internet. We look forward to your comments on reCAPTCHA, where you share your thoughts and how it might come to be in the future.

Resources: 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9