BBC reports how researchers at Carnegie Mellon University have put CAPTCHAs to good use. (CAPTCHAs are the characters you have to type in to demonstrate you're not a robot, for example when posting comments on Blogger.)

Instead of presenting an image of random characters to transcribe, the CMU researchers are presenting small pieces of old books in need of being digitized. The pieces presented as CAPTCHAs are those which optical recognition techniques don't recognize. They compare responses from different people and thus a whole book can be digitized. Inventive!

I hope they build a useful and solid API that can be used to do this on Internet sites worldwide.