Why do we always need to prove that we are human?
A team of researchers at Carnegie Mellon University developed CAPTCHA, which stands for “Completely Automated Public Turing Test to Tell Computers and Humans Apart,” in the early 2000s. For the first time, CAPTCHA meaningfully differentiates between human and bot activity online. This advancement has curbed automated scourges of the Internet, such as the mass creation of spam email accounts and fraudulent responses in online polls, and they are essential tools that make the modern Internet still usable today. Continuing.
So I decided to set myself a challenge. It’s refusing to complete the CAPTCHA because I refuse to prove my humanity to the computer.
But John Langford, a computer scientist who helped create CAPTCHA as a graduate student 20 years ago, is surprised that CAPTCHA has survived this long. “I was hopeful that machine learning would eventually succeed in making his CAPTCHA unnecessary,” he told me.In an interview. “But that hasn’t completely happened yet.”
CAPTCHAs aren’t going away; they’re becoming more complex and more prevalent. It seems that on the Internet he cannot spend a day without solving something.
So I decided to set myself a challenge. It’s refusing to complete the CAPTCHA because I refuse to prove my humanity to the computer. I knew the purpose of CAPTCHAs was noble, to improve the Internet experience by reducing spam and preventing bots from hoarding concert tickets. Still, I wanted to try it. In the meantime, I tried to figure out why her CAPTCHA still exists and what its evolution suggests about the future of the Internet. As I continued reading the story, I began to see another wrinkle in how labor and AI interact.
History of CAPTCHA
Andriy Sherenkov/Getty
In the early 2000s, the Internet had problems. As spammers’ abilities have improved, they have been able to create programs that allow them to create countless free e-mail accounts on services such as Yahoo!. You will receive an email within seconds. This led to an explosion in spam.
A team of Carnegie Mellon University graduate students Louis von Ahn, John Langford, Nicholas Hopper, and their advisor Manuel Blum set out to find a solution. The group recognized that there needed to be some way to differentiate between humans and computers online. However, the test should be able to be solved like this: every They are humans, and the success rate of computers is low.
In the end, the team settled on text recognition. They distort the image of a word and ask the user to identify it. This worked much better than previous experiments. Computers were not good at reading distorted text. Humans, on the other hand, were much better at identifying the letters present, even if they didn’t know the meaning of a particular word. You don’t even need literacy to solve CAPTCHAs. All you have to do is match the characters on the screen with the characters on the keyboard.
This test was conducted at Yahoo! Email was soon used millions of times each day. But in his subsequent decade, a few things happened. First, Google purchased the latest version of a technology called reCAPTCHA to digitize vast amounts of old text. By providing each user with two words (one artificially distorted and one from an old New York Times article), the computer takes advantage of the unwitting human participants. transcribe those articles. But secondly, computers have become better at identifying distorted text. According to his internal Google research in 2014, AI can now read his most distorted CAPTCHAs with 99.8% accuracy. On the other hand, what about humans? Only 33%.
According to John Langford, this was always part of the plan. CAPTCHAs are designed to be a win-win situation. In other words, you can either make your computer inaccessible, or you can enable it to solve problems that it has not been able to solve before. CAPTCHAs allow computers to read distorted text.
So the next development was to replace the text with images. Google reCAPTCHA may ask you to identify the boundaries of your motorcycle by clicking on the square where it resides.
These new CAPTCHAs appear to be more complex and less accurate. I once came across a test that asked me to identify things in an image that didn’t contain either a bus or a car. Also, understandably, I was wondering whether I should click on the square that contained only a small slice of the edge of the traffic light. I’m not alone in this feeling. There are a lot of internet users on Twitter complaining about his CAPTCHAs being confusing or simply wrong.
get a third world job
Artem Stepanov/Getty
My self-imposed CAPTCHA ban has prevented me from applying for jobs. I told myself that if they couldn’t even believe I was a real person, then that wasn’t the kind of company I wanted to work for. It also stopped me from fully checking my astrology reports, which was probably for the best, too. Is stubbornness the root of my rising sign? You never know.
In the end, I decided I didn’t need to log into my Airbnb account. I have given up on the desire to pre-pay for movie tickets. While watching a soccer game using a VPN, it says “abnormal traffic detected” from my boyfriend’s IP address and I can’t use Google search.
But I decided it was time to dig into the cheap labor that keeps spammers going, so I stopped self-policing.
Despite the “low value” of email addresses and other Internet features protected by CAPTCHAs, a CAPTCHA-cracking industry of unclear size still exists. A quick search on Google will reveal several websites that promise cheap and fast CAPTCHA resolution for a nominal fee. These websites (“CAPTCHA farms”) offer to solve 1,000 reCAPTCHAs for about $3. On the other hand, a 1,000 text CAPTCHA costs just $1.
So, after weeks of refusing CAPTCHAs, I decided to break the ban and give this “sure-fire way to earn extra income on the internet” a try. [sic]” says 2Captcha. Additionally, if this writing task doesn’t go well, it’s a good idea to have a backup.
I signed up for an account on Kolotibablo and solved a basic text recognition CAPTCHA within minutes. (To resolve reCAPTCHA, I had to download root access software to my computer, but that idea didn’t excite me.)
These five CAPTCHAs were the only ones that took about 10 minutes to solve in total, due to low demand for the basic CAPTCHAs that were allowed before being banned by entering an additional ‘2’. At least someone paid me a little money to prove my humanity, I thought, even if I couldn’t prove it continuously or for a very long time. Ta. I blamed my failure on lack of practice.
But who spends every penny solving thousands of CAPTCHAs, spending hours entering numbers, and identifying traffic lights? Perhaps unsurprisingly, these companies We found ourselves relying on labor from some of the most economically depressed regions of the world. About a quarter of the site Anti-Captcha’s employees are from Venezuela, according to the site’s own data. Indonesia, Vietnam, India, Pakistan, Philippines and Ukraine top the list. The companies say their workers earn between 25 and 80 cents an hour.
These jobs allow you to work from anywhere with an internet connection, using just your smartphone or computer. And certainly, in many parts of the world, $2 to $4 per day can be even more expensive than in the United States. But it is also clear that, despite these companies claiming to provide easy employment for virtually everyone, they have little regard for the well-being of their workers. I joined his three Facebook groups for CAPTCHA solvers around the world. Each group has several thousand members. These posts have been flooded with complaints from workers who say they have been banned from the platform for unclear reasons.
With the exception of Death by Captcha, most of the companies that solve CAPTCHAs did not respond to my requests for comment. The company advertises its ability to solve many common types of CAPTCHAs, as well as its ability to help people around the world solve CAPTCHAs for them, stating that “anyone who wants to work can solve CAPTCHAs for us. We can resolve the issue,” he said, then declined further questions about his location or revenue. Those workers. He is the highest earning Polish user in the world on Kolotibablo in the last 7 days, during which time he has solved over 106,000 reCAPTCHAs. The user earned a total of $110.45.
CAPTCHA Capture
The future of CAPTCHAs may look different than the image and text questions we’re used to. John Langford believes that the result of a manual CAPTCHA test is probably just one of several signals that providers use to determine humanity, and that it’s probably not a very important signal. he said.
Langford believes CAPTCHA is worth the sacrifice. Yes, they are annoying. And he said he failed the CAPTCHA just like the rest of us. But “to me, CAPTCHAs seem like a necessary evil,” he says. “Either you pay for your account or there’s some kind of barrier to full automation. … I think it would be nice to be able to give people something.” [for free] To help them get through the day. ”
However, given the prevalence of CAPTCHA farms and the potential for continued advances in machine learning, it is unclear how long this effect will last. Perhaps the main value right now is getting millions of answers to the questions companies choose (Where is the traffic light? What does a spoon look like?) that will influence the composition of future AI products. It’s ability.
CAPTCHA farms have been around for at least a decade. But they are part of a broader trend of using low-wage labor to solve technology problems. As we step into a brave new world where AI is rampant, I wonder if there will ever be a day when technology facilitates the availability of cheap labor, rather than fostering a fully autonomous future. I think so. Until then, it’s back to working on CAPTCHAs. However, it is to prove my humanity.





