As if CAPTCHAs don't already seem devious enough in their attempts to weed human visitors from robo-visitors, Google has come up with a new way to utilize the technology to gather data from guesses visitors submit. Instead of offering two jumbled or blurry words, ReCAPTCHA boxes (upgraded CAPTCHAs) are starting to show users a real image alongside a jumbled word. The image contains a legible but perhaps difficult to scan (via OCR) address sign or street sign from Google StreetView. When a visitor sees the street sign and identifies it, Google pops this information back into StreetView (assuming it aligns with a guess it made beforehand as to what the sign said) in order to better align addresses with images and coordinates. Google confirmed to TechCrunch that the new ReCAPTCHAs are part of a new experiment it's conducting.
Google has previously utilized CAPTCHA boxes to let humans decipher words from old books for the Google Books project, as well as old print newspapers for the Google News archives, but has never used this method with images. It's a cool new way to collect data while screening web visitors, essentially turning people into bots in the process.