In 1770, Wolfgang von Kempelen created what was perhaps the world’s earliest innovation in the field of artificial intelligence: an automaton known as the Mechanical Turk that was capable of defeating all but the most skilled of chess players of the time. It was a mechanical marvel, and Kempelen would give demonstrations for some of Europe’s most influential leaders, allowing observers to view the inside of the Turk through a half dozen or so doors all over the device, which comprised a table and an attached detailed, life-sized Turkish chess player.

There was a secret, though. The device was controlled by a chess master, hidden away in a secret area inside the table. Hence the title, “artificial artificial intelligence.”

I think this teaches an important lesson about computing: sometimes the easiest way for a computer to accomplish a task is to have a human inside the box. This lesson was well learned by Amazon.com; in 2005, they launched the Amazon Mechanical Turk, which allows programmers to automatically and nearly transparently outsource “human intelligence tasks” to the Turk.

There are several other projects which use humans to accomplish tasks that computers cannot. One that I’ve mentioned before is the ESP Game, which is a surprisingly addictive game in which two randomly paired, anonymous players try to agree on tags to describe an image. The lofty goal is to label all of the internet’s images so that they can be sorted and searched by computer: a computer-centered task that, as it turns out, works best if there’s a human inside the box.

Another great project, one that I was remiss in excluding from my CAPTCHA post, is called reCAPTCHA. It’s a pretty standard-looking text CAPTCHA, with a twist: there are two words. One of these words is an unknown word scanned from a piece of literature; the other word is known by the computer. The result is a CAPTCHA that puts a human inside of the computerized box dedicated to digitally preserving our literary heritage. A laudable use of an otherwise irritating web hurdle.

This brings up the concept of “crowdsourcing”, a concept that describes the use of large groups of people who, together, can accomplish tasks that are individually (and sometimes, computationally) insurmountable. reCAPTCHA is an example of this: a single person can hardly be expected to transcribe a classic work of literature, but if everyone were to occasionally recognize a single word while carrying out a task they would do anyway (passing a CAPTCHA), then everybody wins.

By the way, you can check out a crowdsourced art project at http://www.thesheepmarket.com. Using the Amazon Mechanical Turk, these folks collected 10,000 simple drawings of sheep. I love it.

I wonder if there’s a way to use the Vidoop grid for this sort of thing. I doubt a communal art project would be possible, but what if we could use it for image tagging? Let’s say that, every once in a while, an uncategorized image is thrown into your grid, in addition to your normal images. Then, sometimes, you might choose it instead of one of one of your images — then the image you neglected to choose has the same category as that uncategorized image.

Well, that’s something, but it’s not anything particularly great. It would be awesome if we could figure out a way to use the image grid to crowdsource some kind of project to improve the world like reCAPTCHA does with their human-proving.