Surely every day while
browsing, you are exposed to a quiz that asks "Are you a robot?" On a
site or in a specific application, and it is sometimes annoying after you spend
time entering the necessary information and reach the last step to accomplish
what you want, this message appears in front of you with a specific test
through which you prove that you are human and not a robot, and while the
matter is not new It's been known for years, but you certainly wondered why
this message was and why I need to prove that I'm human and not a robot? Let me
tell you the answer through the lines of this article and I'll give you the
reason behind the process.
The story behind the test
The idea of proving
that you are human is actually a side effect of almost everything in life
reliant on a computer of some kind, and with the development of devices and the
possibility of automating many systems and processes, “bots” appeared as
shorthand for robots, and the idea behind it is that these bots perform
multiple operations instead of people in the least amount of time. For example,
a person thinking about mining cryptocurrencies can prepare a code that is a
“bot” or a robot that searches online shopping sites for special-type graphic
processors that can be used for the mining process. This bot works around the
clock to complete more than one purchase of processors Graphics required in the
least amount of time before being carried out from the market.
So, on the one hand, you
find that there is a benefit to these bots in terms of shortening time and
reducing human intervention, but on the other hand, this topic may be exploited
to collect data or perform any illegal activity such as generating followers in
huge numbers or voting on a particular survey, and thus, as a result, it was
invented In the 1990s, the so-called CAPTCHA codes were used to determine
whether you were a human or a robot.
Many sites need to
determine who is dealing with them to avoid problems such as hacking, for
example, where it is confirmed whether the person logging in is a human or a
robot, as well as when adding comments on a specific matter to avoid false
comments and to ensure that they stem from a human. For example, examples of
activities that may be used to achieve certain goals, which may sometimes be
harmful.
These tests took many
forms, and you find that some of them require writing some letters and numbers
written in a way that is interspersed with certain distortions such as fog, or
they are written in a zigzag way and connected to each other and interspersed
with many lines, and the aim behind this is to deceive the algorithms that
recognize the letters automatically. OCR can be used to optically recognize
letters and numbers and be able to extract them correctly from images, so if
the test is regular letters and numbers (not jagged or distorted) a robot can
easily bypass it.
Tests to train artificial intelligence
The situation remained
the same until 2007 when a project appeared under the name reCAPTCHA and the
idea was to redesign the verification test to perform other meaningful tasks
such as training artificial intelligence systems. This project was developed by
computer science scientists at Carnegie Mellon University but appeared to the
public in 2009 The year Google acquired the project and began using it to
develop its map data.
The idea here is that
the user will be asked to identify certain objects such as a street, plane, or
traffic light on a road, and the images used in the test are extracted from
Google Street View. But some may not know that solving this test helps Google
train its AI systems. As I mentioned, when the project was developed, the tests
were redesigned to perform meaningful tasks later, and the irony here is that
by pointing objects in the images to confirm that you are human, you help
machine learning tools identify the contents of the different images, which in
a way helps improve Google products.
It is possible by
identifying different traffic lights, bridges, and intersections, you help
train AI models to develop self-driving cars in the future, even though Google
confirmed in 2019 that the data for these tests will not be used for such
purposes.
The exploitation of
CAPTCHA tests to develop artificial intelligence systems or to perform side
tasks related to the reCAPTCHA project, but appears in more than one test. If
the test is about identifying things from a set of images, then you are here to
help improve artificial intelligence systems, but if the test requires writing
certain words and phrases, here it is about archiving books to keep them for
future generations.
This might sound a bit
shocking but actually, instead of hiring someone to manually enter or check the
content of a textbook, captchas were used that ask the user to type in the
words for that task. As of 2011, the Recaptcha tests were able to digitize the
entire Google Books archive - as well as 13 million New York Times articles
dating back as far as 1851.
So, in short, these tests only started to make sure that the person using a site is human and not a bot, and later took advantage of it to achieve other goals and entrap us in it without our knowledge.