A Dartmouth compsci prof, Dan Rockmore is up to some interesting stuff.
He's been designing computer software that basically takes hi-res images of paintings, turns the visual information into quantifiable data, and compares that information with other data sets from other paintings in order to determine authenticity.
This is sure to cause a firestorm in the art history world if it works. So many paintings that are attributed now to a master were probably accomplished (mostly) by their students or workshops. So questions like "how many brushstrokes does Rembrandt have to paint himself to constitute 'a Rembrandt?'" become a lot more problematic and more transparently subjective. It will certainly force people to reexamine the cult of the master artist and perhaps even result in a new criterion for greatness. This could affect the pricing, insurance, and study of art more than anything since, perhaps, Vasari.
First, the damn thing has to work, though.