No. 3043: CAPTCHA

by Andy Boyd

Click here for audio of Episode 3043

Today, are you human? The University of Houston presents this series about the machines that make our civilization run, and the people whose ingenuity created them.

In 1999, an online poll was released asking what school had the best program in computer science. As you might guess, it was an invitation for creative hacking. Students at both Carnegie Mellon and MIT set loose bots on the internet that cast votes for their respective schools. Both wound up with twenty times more votes than their closest competitors. MIT won by the slimmest of margins.

Bots that prowl the web impersonating humans are more than a nuisance; they can gum up the operation of legitimate businesses. Imagine going to purchase tickets for a basketball game the moment they’re released for sale, only to find they’re all gone – purchased by fast-acting bots.

To counter such activity, many websites make use of CAPTCHA – those often frustrating boxes with blurred letters we’re required to decipher. While annoying, their story’s a telling tale about humans and artificial intelligence.

Simple captcha that contains the letters “smwm”. Photo Credit: Wikimedia Commons

The story is wonderfully captured in the very acronym CAPTCHA: Completely Automated Public Turing test to tell Computers and Humans Apart. And one of the best ways to tell humans and computers apart? Find something we’re good at that computers aren’t. Rather than feeling annoyed, we should raise our hands in triumph when we decipher those wavy letters. We can do it. Computers can’t.

The idea behind CAPTCHA isn’t limited to misshapen letters. We might be shown a collection of animal pictures including hamsters, dogs, and cats, then asked to pick out the cats. Not hard for humans. All but impossible for computers. Or suppose you were asked to type spoken words heard on your computer’s speakers. Our brains are wired to hear human speech even in the presence of all sorts of noise. Computers fail when it’s deathly quiet.

A captcha that uses photos instead of words. Photo Credit: Andy Boyd

Still, for all their limitations, computers are always getting better at what they do – thanks, of course, to the help of talented humans. If those letters in the box seem like they’re getting harder and harder to decode, they are. As bots for deciphering improve, we need to rely more heavily on our human proclivity for sorting things out.

Example of a simple captcha. Photo Credit: Andy Boyd

CAPTCHA Tweet is an application for posting tweets as CAPTCHA. Photo Credit: Flickr/ Shinseungback Kimyonghun

But not to despair. Even as bots develop more humanlike capabilities, computer scientists are coming to our aid. Rather than finding more difficult tests to determine if John Smith is actually human, artificial intelligence can be used to analyze Mr. Smith’s activity on the website. If it’s determined he’s acting more like a human than a computer, he’s given a pass. No need to decipher and type – just click and go. The irony is that artificial intelligence is being employed to determine whether or not Mr. Smith is another form of artificial intelligence.

Prove you are not a robot. Photo Credit: Andy Boyd

And what if a determination can’t be made? In that case Mr. Smith will just have to face those letters in a box. It’s still the best test we’ve got. But the story is far from over, and the ongoing battle of bot versus bot will be something to follow.

I’m Andy Boyd at the University of Houston, where we’re interested in the way inventive minds work.

(Theme music)

Are You a Robot? Introducing “No CAPTCHA reCAPTCHA.” From the Google security website: Accessed February 2, 2016.

CAPTCHA. From the Wikipedia website: Accessed February 2, 2016.

CAPTCHA: Telling Computers and Humans Apart Automatically. From the website: Accessed February 2, 2016.

This episode was first aired on February 3, 2016