I’m Not a Robot: Can Captchas Actually Tell?
Artificial intelligence is outsmarting captcha security and it shows
We’ve all seen this little button — some of you may have even tried to click the box out of habit. It pops up on our favorite social media apps, online stores, and news sites, never failing to aggravate us just the slightest with its silly demand that we prove we are human.
Sometimes, it even throws us a puzzle…..
Questioning ourselves, we squint and try to decide if that one box is a bus or just a really big car.
Surely some robot cannot see what we humans see and decipher the right answer to these questions right? Artificial intelligence begs to differ.
“Computers are able to see, hear, and learn. Welcome to the future.” — Dave Walters
Let’s Break Down CAPTCHA’s Battle with Artificial Intelligence
In the early 2000s, a team led by Luis Von Ahn at Carnegie Mellon University developed a random, squiggly letter generator that filtered out bots posing as people.
In 2009, the system was purchased by Google. This was revolutionary and, for some while, the problem of spambots seemed to be solved.
But was it?
You guessed it. Nope. New security measures sparked the desire to break these barriers and a new market of the underground CAPTCHA solving industry rose. These companies hired workers to crack these CAPTCHAs by solving thousands of captcha questions and eventually trained automated ways to breakthrough.
In 2011, Google introduced the new reCAPTCHA which used the digitization of books that were ineligible to be scanned by computer 3. Von Ahn was certain this would be THE solution, telling the New York Times “We’ll be going for a long time.”
A long time ended up being a short time for the new system as AI was trained to be able to correctly answer these CAPTCHA questions and the need for an improved system was present once more.
In 2014, the reCAPTCHA “I’m Not a Robot” we see today was introduced in which a Risk Analysis determines if the user is human simply by their engagement with the box. As simple as it sounds, this method is by far the most complex of CAPTCHA systems yet and being able to crack it is difficult to learn.
Learning is rarely a challenge for artificial intelligence and, ironically, this new reCAPTCHA is the perfect learning environment for AI to learn how to mimic human behaviors to seem human to the analysis.
How does this affect you?
Even if there are successes with limiting bots, these security measures can do little to stop trolls that manipulate social media, fake follower counts, and spread misinformation that can have detrimental effects on our individuality.
On top of the negative impacts on the consumer, CAPTCHAs can negatively impact businesses’ user experiences as well. Studies indicate that the annoyance of having to solve and wait for the box to load can many times lead the user away from your site.
In 2009, Case Henry found that taking out CAPTCHAs could increase customer conversion rates by 3.2%. Is a semi-effective anti-spam protection worth the cost of poor user experience to customers?
What Can Be Done?
At a fundamental level, we want humans interacting with our online platforms and it is imperative to find a solution that makes it difficult and costly for one user to become many. Rather than creating one barrier after the identities are made, the answer is in how we define the identity in the first place.
Creating a new google or Facebook account can take minutes as all you need is a unique email address to become another identity on the internet.
With a tool like humanID, the user needs to use their phone number and coming across more phone numbers means buying more SIM cards which is expensive and difficult to do. humanID is an effective single sign-on service that prevents bots from the beginning in an easy to use way. By integrating a tool like this, businesses can prevent spamming on their platforms while also ensuring a user-friendly implementation.
As AI is evolving, our definition of what it means to be human online must evolve as well. Perhaps CAPTCHA had a moment of glory where it was effective- however brief that may have been. But, the potential for AI is immeasurable and it will only continue to outsmart an inefficient system like CAPTCHA.
In the meantime, moving towards a more sustainable bot prevention system that is built to work against the growth of AI is a promising solution to the problem.
If you like what you’re reading, be sure to applaud this story (did you know that you can hold down the applaud button and it’ll keep adding claps–it’s addictive!) and follow our channel
What’s humanID?
humanID is a new anonymous online identity that blocks bots and social media manipulation. If you care about privacy and protecting free speech, consider supporting humanID at www.human-id.org, and follow us on Twitter & LinkedIn.
