Now That's Using Your Brain

Human Computation

The key to Luis von Ahn's success lies not just in what he does, but in what you do: von Ahn develops techniques that utilize the computational abilities of humans. His innovations have earned him top honors, such as the MacArthur Foundation's "genius grant," a Microsoft New Faculty Fellowship, and a spot on Popular Science's list of 10 Brilliant Scientists of 2006.

As the name suggests, human computation relies on the processing power of people to solve problems that computers cannot yet solve. The potential of human computation peaked von Ahn's interest while he was working on his doctoral degree in computer science at Carnegie Mellon.

"Ten to 20 years ago it was impossible to coordinate more than 100,000 people on one project," he explained, "but the Internet allows 100 million people to work together on the same thing."

And 100 million people are an unparalleled resource.

Because the Internet is becoming increasingly ubiquitous, millions of people are online at any given moment. Even if each one only uses the computer for a few minutes, that time translates to a little bit of processing power—and a little bit of power per person equals a lot of power overall. By filtering that power in a single direction, solutions to real challenges begin to emerge.

reCAPTCHAs

Nearly everyone is familiar with von Ahn's work, if not by name then by function. If you've ever opened a new email account or bought something online then chances are you had to type the correct word in a little box from an image of misshapen letters. By solving the puzzle the computer knows that you're human and not a spammer 'bot.'

The test is called a CAPTCHA (Completely Automated Public Turing Test to Tell Computers and Humans Apart) and computers can't pass it. The first CAPTCHAs were developed at Carnegie Mellon by Luis von Ahn and his Ph.D. advisor Manuel Blum for use by Yahoo! to guard against computers opening email accounts. The program's elegant simplicity has made it the gold standard for Internet security, protecting an estimated 200 million transactions a day.

In addition to email security, CAPTCHAs help ensure the accuracy of online polls, prevent dictionary attacks in password systems, and guard webpages from indexing in search engines.

Following CAPTCHA's introduction, von Ahn discovered a way to improve the program while adding a beneficial component—the deciphered words could help digitize old books for the Internet Archive. Instead of choosing pairs of randomly selected words, reCAPTCHA selects a word misread by optical character recognition (the technology used in digitalization projects) and partners it with a known word. If the user types the known word correctly, then the computer assumes that the unknown word was also typed correctly and places in the digital archive.

According to von Ahn, an assistant professor in the computer science department, over 750 million people (or more than 10% of the world's population) have participated in reCAPTCHAs. That number far eclipses the coordinated efforts some of the world's largest projects from just a quarter-century ago.

The program has been made possible through generous funding from Intel, Novell, the MacArthur Foundation, and ALLADIN/Olympus.

GWAPs

For all of the incredible things that computers can do, there are still some things that humans do better. With that in mind, von Ahn began an ambitious new project to help computers learn by once again tapping into computational abilities of users. This time, von Ahn wanted to improve image search technology.

Whereas reCAPTCHAs succeed because they safeguard computer access, the challenge for von Ahn's latest idea was a bit more taxing—how to get the same number of people involved as were solving reCAPTCHAs. His solution: make the work fun.

By combining the popularity of online gaming with productive tasks, von Ahn created what he calls "games with a purpose," or GWAPs. Not surprisingly, some users spend upwards of 40 hours a week playing the games.

Von Ahn's first GWAP, the ESP Game, displays identical images to a pair of players who describe it by typing in words. When the descriptions match, players earn points and move on to the next image. Matching descriptions help create tags that can be used to improve image search capabilities. And with millions of users online at any give time, the potential for a warehouse of tags is exceptional.

Search giant Google has licensed the technology as Google Image Labeler.

A similar game, Matchin, shows players a pair of identical images and asks them to choose which one they like more. When the choices match, players earn points and are shown a new set of images; at the same time, the computer learns which images users would prefer to see first in a search.

Other GWAPs include Verbosity (word tags), Tag a Tune (music tags), and Squigl (image tracer). Try out the games and see how many points you can earn!

For the Future

Although von Ahn's research makes computers smarter, the real benefit is for the people that use them. As for his next project, von Ahn said he's not sure yet what it will be. "But if [millions of] people are going to be working on a project together," he added, "it had better be something that will benefit all humanity…and the less it feels like work, the more people will participate."

It's no surprise that von Ahn's work has been so well-received. He's collaborated with some of the brightest minds in the field, such as computer science professor Manuel Blum (winner of the A. M. Turing award, the "Nobel Prize" of the computing world, and member of the National Academies of Science and Engineering) and Dr. Josh Benaloh at Microsoft. He also interned at Microsoft Research and the IBM T.J. Watson Research Labs.

"Microsoft has been great," he explained. "They really understand the freedom and openness of research. It was very rewarding working with them."

Whether it's tightening Internet security, archiving the wealth of human knowledge, or improving search engines, von Ahn's work has had a phenomenal impact on the future of computing and the technologies we use everyday. And in the time it takes to type a word, millions of people have helped make that work a success. That type of collaborative spirit is what Carnegie Mellon is known for, and what makes von Ahn's accomplishments so remarkable.