It seems strange that we live in a world where, a few times a week, we're required to enter a series of letters or numbers to prove we're human. But CAPTCHA -- which, no joke, stands for Completely Automated Public Turing Test to Tell Computers and Humans Apart -- is indeed a part of our lives. It's the blurry set (or sets) of numbers and letters that you have to retype to verify that you aren't an automaton. And CAPTCHA uses optical character recognition (OCR) in super stealthy ways.
OCR is a technology that, among other things, is commonly used to convert images of text into editable text. This is important for things like e-books or archived newspapers that are scanned. You want to be able to sort and search the text, and you can't do that unless you've turned the images of characters into coded text characters.
But OCR isn't foolproof, and Google has cleverly appropriated CAPTCHAs to help when OCR can't make out a character or word. When you're answering a CAPTCHA and you see two words, the computer is only authenticating one of them. The other is one that OCR couldn't read when Google was translating an image into text. When you type in the word you see, Google is putting you to work, translating the word OCR couldn't. If enough people type the same word in, it's assumed to be correct. The confirmed word is used in the previously unreadable space, and you've just helped OCR technology get that much better.