Captcha has become an integral part of our life. If you have not encountered it, you are either lucky or you simply do not exist. Small pictures on which it is necessary to mark cars, bicycles, traffic lights, and so on, are found always and everywhere. This is especially true for Google. Do you think they are needed in order to understand if you are a robot? Partly yes, but they also have another secret purpose. We put on a foil hat and read on.
Perhaps what has been said looks a little strange and far-fetched, but you and I have specific facts that make it clear in which place the "Corporation of Good" uses us and converts our free time with you into its own benefit. Personally, this is not pleasant to me and I would not want events to continue to develop like this, but so far everything is exactly like this, and soon it may become even worse. Okay, let's talk about everything in order …
What is captcha
For a better understanding of everything that will be said below, let us first consider the definition of what we used to call such a word as "captcha".
This word is not Russian and in the original is an abbreviation of several English words. It is spelled as a CAPTCHA, and the transcript sounds like "Completely Automated Public Turing test to tell Computers and Humans Apart" (fully automated public Turing test for distinguishing computers and people).
The Turing test is an empirical test, the idea of which was proposed by Alan Turing in his article "Computing Machines and the Mind", published in 1950 in the philosophical journal Mind. His goal was to determine the ability of a machine to mislead a person into thinking that he is communicating with another person.
Promotional video:
It is this principle that underlies the algorithm that determines who is trying to send a request to receive information from the server. Typically, there are three main types of captcha. These are poorly written numbers or letters that a computer cannot determine, mathematical operations that a computer can easily perform, but does not understand that they need to be performed and the definition of objects, so beloved by Google.
How captcha works
We will focus on the third option with pictures. In order to gain access to information, you need to find certain objects in the pictures. For example, bicycles, traffic lights, shop windows and so on.
The usual procedure, you say, and you will be partly right. But have you ever wondered why, with a partially incorrect answer, you are still allowed to access the information you need? It turns out that no one cares if we answer correctly? This can be explained by the fact that the algorithm contains a certain error, which is acceptable when passing this test. Suppose this is true, but why is a simple search engine issue often protected in this way? What is so terrible will happen if a conventional robot finds out how many stars are from the constellation Orion? DDoS protection? Perhaps, but there is also a simpler explanation.
What is AI
Google, like other industry giants, is working on creating its own version of artificial intelligence, or, as it is increasingly called, AI. I am a little skeptical about this phrase, but let's leave this reasoning for another article.
In the modern understanding of the term AI, it represents the ability of a computer to imitate human activities, including through machine learning to learn to identify objects.
As a rule, for such systems to work, they must first be taught by examples. That is, a person shows the bus to the system and says that it is a bus, then shows a fire and says that it is a fire. This continues for some time, after which the system itself tries to determine where what is, and the person says it is right or not.
It is possible that Google's CAPTCHA system aims to do just that. Considering the billions of requests that users around the world send to the server, the training can be very good and complete. Even if you offer a captcha to every tenth user, you will still get a huge amount of data that the company can use to train its systems. At the same time, completely in automatic mode.
Lots of captcha options that we come across every day.
How Google is trying to use us
Of course, this is not officially confirmed, but judging by all the facts, the company is using us. The confirmation is that captcha appears so often and for some reason sometimes misses errors.
If so, just imagine how much money the company is saving on an army of specialists to train the image recognition system. The same applies to those cases when you are invited to read blurry text on the facades of houses. In this situation, the system can be trained to determine house numbers when scanning streets for various map services.
Captcha with house numbers.
For example, the picture above shows several examples of captcha with house numbers. On the one hand, this is just a not very contrasting image that the robot most likely will not be able to recognize. On the other hand, why exactly house numbers? House numbers are gold for Google maps. The creators of the service will not need to arrange the houses separately. The system will be able to figure out where which number is and put it on the map in automatic mode. And then we will also check the result, when we do not find the desired house and send the edits for consideration, having received the ephemeral status of a member of the Google team as a reward. After all, house numbers are not drawn, but actually photographed, and most likely by Google cars, which roam at full speed along the roads, filming streets for "panoramas".
This will help autonomous cars work in the future. To confirm this, it is worth remembering captcha pictures with traffic lights, pedestrian crossings and intersections. It is these pictures in not the best quality that the autopilot cameras will produce and the system will need to recognize them. So far, there are no autonomous machines in the broadest sense, but data for them is already accumulating.
Captcha with the search for elements of road infrastructure.
Additional evidence that it is Google who is trying to pump up its services with information is the Chinese identification systems, which boil down to a simple operation, such as placing a checkmark in the right place or swiping across the screen. Google doesn't work for them and they don't have to do this kind of data collection.
An example of Chinese captcha.
Another example of Chinese captcha.
In addition to pictures, Google offers to recognize voice text. Do you understand what I mean? Google has its own assistant, who for learning needs to communicate with someone and not only listen to requests, but also understand how well a person understands what he was told. For this, the voice captcha system is used.
If so, is Google making us work for itself? It turns out like this! Imagine how each of us is contributing to Google's great AI learning experience, all for free. Personally, I find it a little annoying after every Google request to solve pictures, but I noticed that this goes from one user to another. At one time, no one understood me when I said that almost every search query from a smartphone for me is accompanied by captchas.
It turns out that the captcha epidemic wanders from user to user. In my opinion, this once again confirms the theory given above.
Do you think this could actually be? Write in the comments and take part in the poll below.
Artem Sutyagin