You Have Been Helping Companies Improve AI Unknowingly

By Dennis Hertiandi, DIGITS Staff Writer

For years, big technology companies have tried to improve their Artificial Intelligence (AI) system. They believe AI is the next step toward better technological environment. But, did you know that you unknowingly help them to improve their AI, and maybe not in an obvious way? This article will dissect shortly what is Ai, what is CAPTCHA, how CAPTCHA improves AI, and other forms that you may or may not assist them in advancing their AI.

What is AI

            According to Merriam-Webster, AI is a branch of computer science dealing with the simulation of intelligent behavior in computers, or the capability of a machine to imitate intelligent human behaviour. AI is sometimes called Machine Intelligence.

            To simplify the definition, AI is how to make machines more human, or to do human like process (like thinking, planning), and to make the human user feels like as if they are interacting with other human, not with a machine or computer. Therefore, even though it is not fully correct, AI is a way to humanize machine.

            Many believes AI’s foundation came from Turing Machine, which was created by Alan Turing back in Second World War to crack enigma code used by German forces. The term of AI itself was first adopted in 1956 by American computer scientist John McCarthy in Dartmouth Conference.

            The whole purpose of AI is to create machine which is intelligent enough to behave and do task intelligently. The problems which AI is facing now is problem solving, knowledge representation, planning, learning, natural language processing, perception, motion and manipulation, social intelligence, general intelligence.

            Although AI might seem to be complicated, some use of it is actually within the grasp of our hand. Many application we use today actually uses AI, like Facebook, Instagram, Google, Apple Siri, Amazon, Twitter, Intel, Microsoft, etc.

            AI is being implemented in daily life as well. For example, the development of automated cars by Google and Tesla, Banking system (maintain book-keeping, invest in stocks, manage property), Video Games, Military, Audit, Advertising, daily task helper, etc.


            CAPTCHA stands for Completely Automated Public Turing test to tell Computers and Humans apart. As the name implies, CAPTCHA is a form of implementation of Turing test, to tell whether the input comes from computer or a human.

            The system was first developed by a team of engineers at Carnegie Mellon University in the early 2000s. The team was led by Luis von Ahn, who wanted to find a way to filter out spambots pretending to be person. They find that using distorted words could be interpreted by humans, but not by computers.

            From distorted words, CAPTCHA improves by using sound, and then they use picture. In recent times, there exists reCAPTCHA, where the words that is used in the CAPTCHA is from a book that Google wants to digitize.

How CAPTCHA improves AI

            We all know what CAPTCHA is. We see them all over the internet. Whether we want to login to a website, access the website, buy tickets, online polls, etc. CAPTCHA can be in form of distorted words, or maybe choose a picture. But, CAPTCHA might be more than that.

            You may realize then when you input a CAPTA

CHA, the word looks like it comes from a book. It usually really is come from a book. Google use reCAPTCHA to actually uses people to digitize their book, without their knowing. By 2011 Recaptcha had finished digitising the entire Google Books archive – as well as 13 million articles from the New York Times back-catalogue dating back to 1851.

            The CAPTCHA and ReCAPTCHA also improves AI ability to read not only typed words, but also distorted words. AI may be smart enough to learn by itself through millions of data in the database and internet, but it also needs to be guided by the human. Some function can only be taught by humans, or maybe they have to replicate the humans doing stuff. That’s why they use CAPTCHA, so the machine can replicate the human work, and learn from it. ACHACh

            You may find this application in Google Translate. In Google Translate, there is a feature where you can take photo and translate the words from the photo. This capability where the machine can translate muddled and distorted words come from the AI learning of words, one of which came from CAPTCHA.

            Now, CAPTCHA uses photos to verify whether you’re a human or a machine. The target is similar, Google and other AI companies wants to teach its AI depth learning, to determine whether the picture is in front or the back. They also want for the machine to differentiate objects, therefore in CAPTCHA you are asked to pick photos with car, bridges, post, sign, etc.

            The capability of image differentiating is useful for several scenarios. First is for their automated car. Their cars need to see lights, other cars, etc. So it is useful for the machine to learn how to differentiate them. The second is to improve Google Maps. And the third is for depth sensor. Mobile phones these days are getting smarter, as it can detect the “depth” of things in the image, which is why they can create bright image with almost no light at all. This is certainly useful as mobile phones market is growing rapidly, and this ability can bring advantage to their phone.

            These are just some example of the usage of depth image. There are so many other advantages that come from image differentiating, depth sensor, etc, and the tools and power to enable the machine to become smarter actually comes from things that nobody cares about such as CAPTCHA.

Other forms

            There are some other forms of data where we unknowingly help companies to enhance their AI. For example, the mannequin challenge. You may think it is a fun challenge, easy to do, etc. But it also benefits Google, as the AI can learn from Youtube’s large database of faces, lots of faces, which enhance their facial recognition software. There’s also exist a website like Amazon’s Mturk where you do basic task and Amazon will give you slight money. For example, type the numbers in the photo, type the address in the photo, pick picture with person, etc. Websites like this actually gains benefits, as the task that you find simple is now hard for AI to replicate, and therefore, by teaching them using your example, it will also enhance and improves the AI’s capability.


            There are a lot of times we are unaware of situation, where we find it is a simple and boring task like CAPTCHA in real life, unknowingly it also improves AI. So, in other words, you are a part of groups who helps to improve AI.

Ray,Shaan. (2018, Aug 11). History of AI. Retrieved from

Artificial Intelligence (n.d) in Merriam-Webster’s collegiate dictionary. Retrieved from

Horowitz.Kate. (2016, June 21). The Surprisingly Devious History of CAPTCHA. Retrieved from

Strickland.Jonathan. (2008, August 8). How CAPTCHA Works. Retrieved from

Dzieza.Josh. (2019, February 1). Why CAPTCHA Have Gotten So Difficult. Retrieved from

O’Malley.James. (2018, January 12). Captcha if you can: how you’ve been training AI for years without realising it. Retrieved from

Burgess.Matt. (2017, October 26). Captcha is dying. This is how it’s being reinvented for the AI age. Retrieved from

Goedegebuure.Dennis. (2016, November 30). You Are Helping Google AI Image Recognition. Retrieved from

Icon made by Freepik from


Leave a Reply

Your email address will not be published. Required fields are marked *